{
  "generated": "2026-10-09T11:30:13+00:00",
  "confirmed": [
    {
      "id": "free-program-fb7cc23ee69b754166b3",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Thirteen free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen prices thirteen models at Free for input and output tokens on its pricing table as of 2026-10-08. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-08: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Union Alpha Free left the free table by 2026-10-08. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free come from providers that state a zero-retention policy, and the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. All other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. For Big Pickle, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free, collected data may be used to improve the model during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free trades heavily discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use your data for model training, and Jev inputs are not used for training.",
        "clock": "All thirteen free models are limited-time promotions, and no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices thirteen models at Free on its pricing table. Twelve are Free for input, output, and cached-read tokens, and Jev 1.13 Free is Free for input and output. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-09: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Step 5 Preview Free, Exo Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a TypeSafe AI System One model that answers typed yes/no, multiple-choice, and rubric questions instead of generating text. Union Alpha Free was no longer listed on 2026-10-09. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, Exo Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to be used to train future Meta models. The providers of Space Bunny Free, LongCat 2.5 Preview Free, and Step 5 Preview Free follow a zero-retention policy and do not use data for model training. Jev prompts and other inputs are not used for training.",
        "clock": "All thirteen are limited-time promotions. No end date is stated for any of them."
      }
    },
    {
      "id": "free-program-e67904f4787639df55b6",
      "program_id": "poolside-laguna-limited-free",
      "event_type": "program_record_corrected",
      "program_name": "Laguna hosted API, free for a limited time",
      "provider_label": "Poolside",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://poolside.ai/models",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for its Laguna models, and the models page says Laguna is free to use for a limited time. On 2026-10-04 that free offer covered both hosted models: Poolside states for Laguna XS 2.1 that free and paid endpoints are both available, with paid use priced at $0.10, $0.20 and $0.05 per 1M input, output and cache-read tokens. No quota for the free endpoints and no end date for the promotion are published.",
        "models_detail": "Poolside-hosted inference served two Laguna models on 2026-10-04. Laguna S 2.1, called with the model ID poolside/laguna-s-2.1, is a 118B total parameter mixture-of-experts model with 8B activated. Laguna XS 2.1 is a 33B total parameter mixture-of-experts model with 3B activated, served at 256K context on Poolside's API. Both are reached through an OpenAI-compatible chat API at https://inference.poolside.ai/v1, which exposes a model listing endpoint and a chat completions endpoint."
      },
      "after": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for its Laguna models. The models page says Laguna is free to use for a limited time and shows Laguna S 2.1 and Laguna XS 2.1 under that offer. The Laguna XS 2.1 launch post says free and paid endpoints are both available, with paid pricing of $0.10, $0.20 and $0.05 per 1M input, output and cache-read tokens. No free-endpoint quota or end date is disclosed.",
        "models_detail": "On 2026-10-09 the models page showed Laguna S 2.1 (poolside/laguna-s-2.1) and Laguna XS 2.1 under the limited-time free offer for the hosted API."
      }
    },
    {
      "id": "free-program-c9d7758250ea6a947c0d",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible organizations in the Launch and Grow usage tiers receive up to 1,000,000 tokens per day in the large-model group and 10,000,000 tokens per day in the small-model group, and eligible Build organizations receive 250,000 and 2,500,000 tokens per day respectively. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed at normal rates. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded.",
        "models_detail": "On 2026-10-08 the 1M/250K group lists gpt-6.1-sol, gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group on the same date lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the first group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible Launch and Grow organizations receive up to 1,000,000 tokens/day in the large-model group and 10,000,000 tokens/day in the small-model group; eligible Build organizations receive 250,000 and 2,500,000 tokens/day respectively. Eligibility is shown on the data sharing settings page, and organizations that do not see the offer there are not eligible. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed normally. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded.",
        "models_detail": "On 2026-10-09, the 1M/250K group listed: gpt-6.1-sol, gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On 2026-10-09, the 10M/2.5M group listed: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also keeps gpt-4.5-preview in its list but marks it shut down, so it is not current access."
      }
    },
    {
      "id": "free-program-c6b7f25b88e00e62adb9",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Accounts with no credit get 10 requests per minute and 60 weighted units per day. Accounts that have topped up get 20 requests per minute and 120 weighted units per day, so a top-up doubles both the per-minute limit and the daily allowance. On 2026-10-07 the free page states that this allowance is shared across the free models for an account, with an additional IP guard, a site-wide free cap of 15 requests per minute, and a limit of 2 concurrent free requests per account. The per-minute limit uses a rolling 60 second window and the daily counter resets at 00:00 UTC. When weighted limits are enabled, longer prompts use more daily units, so long-context requests use the allowance up faster; otherwise each request counts once. The developer docs give the general rule as a separate daily count for each free model and add that a model's own listed allowance governs, which is the shared figure shown on the free page. Past the allowance, requests continue at the paid rate only when the account holds credit, and sending the header X-Free-Fallback: false turns that fallback off; an account with no balance gets 429 until the quota resets. The docs' example rate-limit header (200 requests per minute) is the paid-tier account limit, not the free quota.",
        "models_detail": "Two enabled free-quota models on 2026-10-07: Qwen3.7 Flash (qwen/qwen3.7-flash), with a 1,000,000 token context and text, image and video input, and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), with a 1,048,576 token context and text input. The auto:free route always resolves to a free model. The catalogue changes over time, and only enabled free models are listed."
      },
      "after": {
        "free_terms": "Accounts without credit get 10 requests per minute and 60 weighted units per day on the free models; accounts holding credit get 20 requests per minute and 120 weighted units per day. The daily counter resets at 00:00 UTC, and longer prompts use more units. The free page says these limits are shared across free models, while the docs page says each free model is counted separately. An additional IP limit, a site-wide free cap of 15 requests per minute, and a limit of 2 concurrent free requests per account also apply. Past the quota, requests continue at the paid rate only when the account holds credit and paid fallback is on (a header can turn that fallback off); otherwise the request gets a 429 until the quota resets. The docs' example rate-limit header (200 requests per minute) is the paid-tier account limit, not the free quota.",
        "models_detail": "Two current free-quota models on 2026-10-09 (Qwen: Qwen3.7 Flash and DeepSeek: DeepSeek V4 Flash 0731free) plus auto:free routing; the catalog changes over time."
      }
    },
    {
      "id": "free-program-ab6be87f598dfe42e50c",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 13 free models on 2026-10-08: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. All 13 show Free for both input and output. The documentation page carries a shorter nine-row table that leaves out Ling 3.0 Tiny, Ling 3.1 Flash, Nemotron 3.5 Lightning 30B A3B and Muse Glimmer 30B. Requesty says the models are free for now and that any pricing change will be announced in its changelog ahead of time.",
        "card": "The Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and its free-models page says most major upstream providers do not use API data for training by default. Retention and training policy still differ by the provider serving each model, and Requesty directs users to each model's detail page for the specific policy, so check that page and the underlying provider terms before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 13 free models on 2026-10-09: NVIDIA Nemotron 3 Ultra 550B, Nemotron 3 Super 120B, Nemotron 3 Nano 30B, Nemotron 3 Nano Omni 30B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. The docs page carries a shorter nine-row table. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card is required for the Free plan.",
        "data_treatment": "Requesty says it never uses your data for training. Retention and training policies of the underlying model providers vary by model, and Requesty shows them on each model detail page; check them before sensitive use."
      }
    },
    {
      "id": "free-program-7c98ba805b964fc8ba16",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all reset daily. Models that carry a higher request weight have lower caps, ranging from 50 requests per day and 5 per minute down to 10 per day and 1 per minute.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page, which was updated on 2026-10-07. Nearly all return text, and two return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free). Every free model answers the Chat Completions, Messages, and Responses protocols. The roster changes over time.",
        "card": "No credit card is required for the ten trial calls. Reaching the recurring daily quotas takes a one-time top-up of any amount from $1, and the provider does not state which payment methods it accepts for that top-up."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls, shared across all free models, with no credit card and no expiry. A one-time top-up of at least $1 permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 per minute, and one shared pool of 1M tokens per day, resetting daily. Free models bill $0 within these quotas. Some models carry a higher request weight and have lower caps, ranging down to 10 requests per day and 1 per minute.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page on 2026-10-09. Nearly all are text models, plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free). The roster changes over time.",
        "card": "No credit card is required for signup or the 10 trial calls. The daily free quotas require a one-time top-up of at least $1; the provider does not state which payment methods the top-up accepts."
      }
    },
    {
      "id": "free-program-72065e5ec863d52df811",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "card",
        "data_treatment"
      ],
      "before": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated December 2, 2025, contain no billing or fee language. An Amazon-provided API key is needed to access certain Software (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that Build with Nova is an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication path applies usage charges to an AWS account at Nova Act pricing. No numeric quota, daily limit, or rate limit figure is published on either page.",
        "models_detail": "Amazon Nova models and capabilities are reached through the nova.amazon.com web interface, which the Terms of Use describe as including capabilities to understand and generate text, image, audio, and other content (Terms 1.1). Build with Nova at nova.amazon.com/dev provides the Nova API and lets developers configure Amazon Nova models to create Nova Agents (S-1.1), and Amazon Nova Act is reached with an API key from nova.amazon.com/act. On 2026-10-08 the per-model roster remained behind Amazon sign-in and was not publicly confirmable.",
        "signup": "required; Amazon.com consumer account sign-in, then an API key from the developer surface. The Nova Act user guide states that the API key is tied to your Amazon.com account.",
        "card": "Not required. Amazon documents API key authentication for Nova Act as free of charge with daily limits, and the Nova.amazon.com Terms of Use contain no payment terms. A payment relationship appears only on the separate AWS IAM path, where usage charges apply to an AWS account.",
        "data_treatment": "Amazon records, processes and retains interactions to provide, develop and improve its services, including artificial intelligence models, and may review them manually (Terms 1.2). Information may be stored on servers outside the user's country (Terms 2.1). No opt-out is stated."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (last updated December 2, 2025) contain no billing or fee language, and the Amazon Nova Act user guide states that API key authentication through nova.amazon.com/act is free of charge with daily limits, while the separate AWS IAM route bills usage to an AWS account. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering is intended for experimentation and does not support deploying production scale developer applications (S-1.1). No numeric daily limits are published publicly.",
        "models_detail": "Amazon Nova family on the nova.amazon.com developer surface; page metadata names Amazon Nova 2 Lite, Nova Act, and Nova 2 Sonic. On 2026-10-09 the Amazon Nova Act user guide confirmed that Nova Act can be used with an API key from nova.amazon.com/act. The full per-model roster is behind Amazon sign-in and is not publicly confirmable.",
        "signup": "Required: sign in with an Amazon.com account (users must be at least 18), then get an API key from the developer surface, such as nova.amazon.com/act for Nova Act.",
        "card": "Not stated. Neither the Nova Terms of Use nor the Nova Act user guide says whether a payment card is needed; the API key route is described as free of charge and tied to an Amazon.com account.",
        "data_treatment": "Interactions are recorded, retained, may be manually reviewed, and are used to provide, develop, and improve Amazon's services including artificial intelligence models (Terms 1.2). Information may be stored on servers outside the user's country (Terms 2.1). No opt-out is stated."
      }
    },
    {
      "id": "free-program-6bb5fa04c2e1cee5e811",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-08 the catalog page lists 20 models, and the product page describes more than 20 open-weight models spanning LLM, text-to-speech, speech-to-text, document analysis, and image analysis. The families present on that date are Qwen, Llama, GPT-OSS, Whisper, bge embeddings, Stable Diffusion image generation, and NVIDIA Riva speech. Seven catalog entries are priced at Free outright on 2026-10-08: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for German, Spanish, Italian, and US English. The remaining entries are paid, for example Qwen3.8-27B at 0.4 EUR per 1M input tokens and 2.7 EUR per 1M output tokens. Individual models can be retired under OVHcloud's lifecycle process, and on 2026-10-08 four entries are listed as being decommissioned.",
        "data_treatment": "OVHcloud states that user data is not stored and is never used to train or improve its AI models."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide, per model, across the current AI Endpoints catalog. On 2026-10-09 the public catalog listed 20 models, including Llama, Qwen, GPT-OSS, Qwen3Guard moderation, embedding, image-generation, speech-to-text and text-to-speech models. Models announced for retirement remain reachable through the API for a three-month grace period but are hidden from the catalog. On 2026-10-09 OVHcloud listed Mistral-7B-Instruct-v0.3, Mistral-Nemo-Instruct-2407 and Mistral-Small-3.2-24B-Instruct-2506 as being decommissioned. The authenticated catalog also prices some entries at Free outright: on 2026-10-09 these were the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and four NVIDIA Riva text-to-speech voices. Qwen3.8-27B is paid, at 0.4 EUR per 1M input tokens and 2.7 EUR per 1M output tokens.",
        "data_treatment": "OVHcloud states that user data is never used to train its models and that it keeps only the data strictly necessary for billing. The service is hosted in the European Union, at Gravelines, France."
      }
    },
    {
      "id": "free-program-5823c62365c20b992282",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "models_detail": "On 2026-10-07 the program page names Solar Pro 4 (LLM) and Document Parse as the models included free for approved organizations. Solar Pro 2 and Solar Pro 3 are no longer named in the program, and the API pricing page marks both for end of service on 2026-10-30 (KST).",
        "signup": "Application and approval required. Eligible applicants are K-12 schools and universities, university hospitals, and nonprofit and non-governmental organizations. Applicants must sign up for the Upstage Console with the same email address before submitting the application form, and approved organizations receive detailed instructions by email.",
        "clock": "Applications are accepted through March 31, 2027 at 11:00 PM KST. Approved organizations get free access for up to one year, running from the date of approval until March 31, 2027, so an approval granted late in the window gives less than a full year."
      },
      "after": {
        "models_detail": "Solar Pro 4 (LLM) and Document Parse, as named on the program page on 2026-10-09. Solar Pro 2 and Solar Pro 3 are no longer named in the program, and Upstage lists both for end of service on 2026-10-30 (KST).",
        "signup": "Application and approval required; eligible K-12 schools, universities, university hospitals, nonprofits, and NGOs. Applicants must first sign up for Upstage Console with the same organization email they enter on the application form. After internal review and approval, instructions arrive by email.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Free access lasts up to one year and runs from approval until 2027-03-31."
      }
    },
    {
      "id": "free-program-42afd09c6af42a5ccf1b",
      "program_id": "together-ternary-bonsai-free",
      "event_type": "program_terms_changed",
      "program_name": "Prism-ML/Ternary-Bonsai-27B",
      "provider_label": "Together",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.together.ai/pricing",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "Free model behind a one-time minimum credit purchase. Together AI does not currently offer free trials, and platform access requires a minimum $15 credit purchase for an organization's first purchase, then a $5 minimum for later purchases. The platform is fully prepaid, so a positive credit balance is required and API access is suspended if the balance reaches zero. Ternary Bonsai 27B lists $0.00 input and $0.00 output token prices, so those tokens do not consume the purchased balance under published pricing. Other paid usage can drain the balance. Serverless rate limiting can occur with high request volumes or large bursts of traffic and serverless performance is best effort, so this is not a guaranteed unlimited allowance.",
        "card": "A payment method and a minimum credit purchase are required: $15 for an organization's first purchase and $5 for purchases after that. A card is not the only documented payment route, because Together also documents US bank account ACH purchases, which deposit credits only after the payment clears. A credit or debit card is required for auto-recharge."
      },
      "after": {
        "free_terms": "Free model after a one-time minimum $15 first credit purchase (later top-ups have a $5 minimum), as stated by Together on 2026-10-09; a positive balance is required. Ternary Bonsai 27B lists $0 input and $0 output token prices, so its tokens do not consume the purchased balance under published pricing. Other paid usage can drain the balance, and API access is suspended at zero until credits are added. Rate limits are dynamic and best-effort, not a guaranteed unlimited allowance.",
        "card": "A payment method and a minimum $15 first credit purchase are required; later purchases have a $5 minimum. A card is not the only documented payment route: Together also documents US bank account (ACH) payments, while a credit or debit card is required for auto-recharge."
      }
    },
    {
      "id": "free-program-339618652a875590ea01",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction and classification pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not supported on the Lite plan.",
        "models_detail": "Ready-to-use IBM and third-party foundation models can be prompted in the Prompt Lab or programmatically from the Lite instance, subject to regional and current catalog availability. On 2026-10-08 IBM's supported foundation models documentation listed the IBM and third-party models provided with watsonx.ai for pay-per-token inferencing, and listed deploy-on-demand models separately, where hosting and access fees are billed by the hour. Custom foundation models are not supported on the Lite plan. An exact Lite-specific model roster is not stated."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 pages per month of document text classification and extraction, and a plan-level limit of 2 inference requests per second. Custom foundation models are not available on Lite.",
        "models_detail": "Ready-to-use IBM and third-party foundation models available to the Lite instance, subject to regional and current catalog availability. IBM's supported-model surface describes IBM and third-party models; custom foundation models are not available on Lite. An exact immutable Lite model roster is not stated."
      }
    },
    {
      "id": "free-program-2b80c013540b4c6b94ef",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-10-08: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 is outside the free set at $0.09 in and $0.36 out per 1M, and Tencent Hy3 is outside it at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms say that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Ten catalog entries listed at $0.00/1M on 2026-10-09: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante; inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; StepFun Step 5 Preview; Upstage Solar Mini 4. On 2026-10-09 Upstage Solar Pro 4 was no longer listed at $0.00 (now $0.09 in / $0.36 out per 1M), and Tencent Hy3 remained paid at $0.13 in / $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, and may make Your Data and Service Data available to third parties for their own use. The terms also say that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support, except as described in the Privacy Policy. The plan page does not say whether Privacy Mode is available on the Free plan."
      }
    },
    {
      "id": "free-program-2517cd57026b714ac5a9",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-09T00:00:00+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (13 models on 2026-10-08) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee, and once the rate limit is reached Kenari returns HTTP 429 and advises waiting for the Retry-After header when the response carries one.",
        "models_detail": "Thirteen provider-hosted :free text models per the Solo card on Kenari's homepage on 2026-10-08 (the same card read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog). Kenari's quickstart notes that the free-model list changes over time and points to a public models endpoint for the list that is active now."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (14 models on 2026-10-09) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Fourteen provider-hosted :free text models per Kenari's homepage on 2026-10-09 (it read seventeen on 2026-09-17, eleven on 2026-09-04 and twelve on 2026-08-27); Kenari says the free roster changes over time with the provider catalog."
      }
    },
    {
      "id": "free-program-f5e660246c9e3b90e8ce",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-07: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. Several of the free models appear a second time in the catalog as separate paid entries at list prices.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms state that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-10-08: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 is outside the free set at $0.09 in and $0.36 out per 1M, and Tencent Hy3 is outside it at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms say that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-f270edafacdb498daa67",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "MythoLite is priced FREE for both input and output tokens on Mancer's model table, so it can be used without buying credits. Mancer's pricing page says it offers free models to test with before making a purchase, and that free models can still be used when an account's credit balance is negative. The same page states that some models are free while most have a per token credit cost, so the free roster is not necessarily limited to MythoLite. Mancer provides no interface of its own, so the free models are reached through its API from another application that supports it.",
        "models_detail": "MythoLite, a LLaMA 2 13B model tagged Creative and Roleplay, carried a FREE price for both input and output tokens on Mancer's model table on 2026-10-07. The table does not state its context length or its maximum output. The other LLaMA 2 13B model listed, MythoMax, was priced at 0.14 credits per input token and 0.24 credits per output token on the same date."
      },
      "after": {
        "free_terms": "Mancer's live model table shows MythoLite with an input token cost and an output token cost of FREE. Mancer's prices page says that some models are free while most have a per-token credit cost, and it tells users to test on the free models before making a purchase. Free models stay usable even when an account's credit total is negative, so free usage does not depend on buying or holding credits.",
        "models_detail": "MythoLite is the only model marked FREE for both input and output tokens on Mancer's model table as read on 2026-10-08. It is listed as a LLaMA 2 architecture model at 13B parameters, tagged Creative and Roleplay, and shown as online. Mancer does not state MythoLite's context window or maximum output length. Paid models on the same table include MythoMax, listed at 0.14 credits per input token and 0.24 credits per output token."
      }
    },
    {
      "id": "free-program-dbedb96354752ac413dc",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free models page listed 13 free models on 2026-10-07: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. Each is priced at Free for input and output. The documentation page carries a shorter nine row table that does not include Ling 3.0 Tiny, Ling 3.1 Flash, Muse Glimmer 30B or Nemotron 3.5 Lightning. Requesty says the models are free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training, and that most major providers do not use API data for training by default, with each model detail page carrying that provider's retention and training policy. Because free routes run through different model providers, treatment is not stated uniformly; check Requesty and the underlying provider policies before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 13 free models on 2026-10-08: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. All 13 show Free for both input and output. The documentation page carries a shorter nine-row table that leaves out Ling 3.0 Tiny, Ling 3.1 Flash, Nemotron 3.5 Lightning 30B A3B and Muse Glimmer 30B. Requesty says the models are free for now and that any pricing change will be announced in its changelog ahead of time.",
        "data_treatment": "Requesty says it never uses customer data for training, and its free-models page says most major upstream providers do not use API data for training by default. Retention and training policy still differ by the provider serving each model, and Requesty directs users to each model's detail page for the specific policy, so check that page and the underlying provider terms before sensitive use."
      }
    },
    {
      "id": "free-program-d51beb03bb3114b219e2",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "clock"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), and an API key is tied to the holder's Amazon.com account. As documented on 2026-10-07, the AWS Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication method instead applies usage charges to an AWS account. The Terms add that usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the offering does not support deploying production scale Developer Offerings (S-1.1). Amazon publishes no numeric daily limit, quota, or rate limit for the free route.",
        "models_detail": "Amazon Nova models and capabilities on nova.amazon.com. The Nova Act developer guide names Nova Act as reachable with a free API key. On 2026-09-17, the public pages did not publish a per-model roster.",
        "card": "not required; the Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and the Nova Terms of Use contain no payment terms. The paid alternative is AWS IAM authentication, which bills usage to an AWS account.",
        "clock": "No end date is stated. Amazon may change, suspend or discontinue Amazon Nova at any time without notice (Terms 2.3)."
      },
      "after": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated December 2, 2025, contain no billing or fee language. An Amazon-provided API key is needed to access certain Software (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that Build with Nova is an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication path applies usage charges to an AWS account at Nova Act pricing. No numeric quota, daily limit, or rate limit figure is published on either page.",
        "models_detail": "Amazon Nova models and capabilities are reached through the nova.amazon.com web interface, which the Terms of Use describe as including capabilities to understand and generate text, image, audio, and other content (Terms 1.1). Build with Nova at nova.amazon.com/dev provides the Nova API and lets developers configure Amazon Nova models to create Nova Agents (S-1.1), and Amazon Nova Act is reached with an API key from nova.amazon.com/act. On 2026-10-08 the per-model roster remained behind Amazon sign-in and was not publicly confirmable.",
        "card": "Not required. Amazon documents API key authentication for Nova Act as free of charge with daily limits, and the Nova.amazon.com Terms of Use contain no payment terms. A payment relationship appears only on the separate AWS IAM path, where usage charges apply to an AWS account.",
        "clock": "Standing developer offering. The Nova.amazon.com Terms of Use were last updated December 2, 2025, and Amazon states it may change, suspend, or discontinue Amazon Nova, or any part of it, at any time without notice. No end date is stated."
      }
    },
    {
      "id": "free-program-bb67342dac12062b6a0a",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible Launch and Grow organizations receive up to 1,000,000 tokens per day in the large-model group and 10,000,000 tokens per day in the small-model group. Eligible Build organizations receive up to 250,000 and 2,500,000 tokens per day respectively. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed at normal rates. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded. Eligibility is shown on the data sharing settings page, and an organization that does not see the offer there is not eligible.",
        "models_detail": "On 2026-10-07 the 1M/250K group was: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group was: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The larger quota in each pair applies to Launch and Grow organizations and the smaller to Build organizations. The page also retains gpt-4.5-preview-2025-02-27 in the first group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible organizations in the Launch and Grow usage tiers receive up to 1,000,000 tokens per day in the large-model group and 10,000,000 tokens per day in the small-model group, and eligible Build organizations receive 250,000 and 2,500,000 tokens per day respectively. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed at normal rates. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded.",
        "models_detail": "On 2026-10-08 the 1M/250K group lists gpt-6.1-sol, gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group on the same date lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the first group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-b75bbecc04b13b727e26",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Thirteen free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen prices thirteen models at Free on its pricing table as of 2026-10-07, free for input and output tokens, and free for cached reads on every one of them except Jev 1.13 Free, which lists no cached read price. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-07: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model called through the systemone endpoint rather than a text generation model. Union Alpha Free left the free table by 2026-10-07, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free may use collected data to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints whose use is logged for security and to improve NVIDIA products and services, and users are told not to submit personal or confidential data. Muse Spark 1.3 Contributor Free carries permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use the data for training, and prompts sent to the Jev endpoints behind Jev 1.13 Free are not used for training.",
        "clock": "All thirteen free models are described as limited-time promotions; no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen prices thirteen models at Free for input and output tokens on its pricing table as of 2026-10-08. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-08: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Union Alpha Free left the free table by 2026-10-08. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free come from providers that state a zero-retention policy, and the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. All other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. For Big Pickle, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free, collected data may be used to improve the model during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free trades heavily discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use your data for model training, and Jev inputs are not used for training.",
        "clock": "All thirteen free models are limited-time promotions, and no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-b4c18e467e12831a812a",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, and 100 text-extraction pages per month, which IBM's pricing page states as 100 documents per month. The plan carries a limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan.",
        "models_detail": "Ready-to-use IBM and third-party foundation models are available for per-token inferencing, subject to regional and current catalog availability. On 2026-09-24 the IBM models provided for inferencing were granite-4-h-small, granite-guardian-3-8b, and granite-8b-code-instruct, with the last two marked deprecated, and the third-party models provided for inferencing came from OpenAI, Meta, Mistral AI, and BigScience. IBM does not publish a separate model roster for the Lite plan, and custom foundation models are not supported on it. Deploy-on-demand models are billed by the hour."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction and classification pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not supported on the Lite plan.",
        "models_detail": "Ready-to-use IBM and third-party foundation models can be prompted in the Prompt Lab or programmatically from the Lite instance, subject to regional and current catalog availability. On 2026-10-08 IBM's supported foundation models documentation listed the IBM and third-party models provided with watsonx.ai for pay-per-token inferencing, and listed deploy-on-demand models separately, where hosting and access fees are billed by the hour. Custom foundation models are not supported on the Lite plan. An exact Lite-specific model roster is not stated."
      }
    },
    {
      "id": "free-program-9ed556e2bc6137a4854d",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. OVHcloud describes the offer as more than 20 open-weight models, and the catalogue listed 20 models on 2026-10-07, covering Llama, Qwen, Qwen3Guard, GPT-OSS, bge embeddings, Stable Diffusion image generation, Whisper speech to text, and NVIDIA Riva text to speech. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalogue also prices a few entries at Free outright, on 2026-10-07 these were the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text to speech voices, while Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-08 the catalog page lists 20 models, and the product page describes more than 20 open-weight models spanning LLM, text-to-speech, speech-to-text, document analysis, and image analysis. The families present on that date are Qwen, Llama, GPT-OSS, Whisper, bge embeddings, Stable Diffusion image generation, and NVIDIA Riva speech. Seven catalog entries are priced at Free outright on 2026-10-08: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for German, Spanish, Italian, and US English. The remaining entries are paid, for example Qwen3.8-27B at 0.4 EUR per 1M input tokens and 2.7 EUR per 1M output tokens. Individual models can be retired under OVHcloud's lifecycle process, and on 2026-10-08 four entries are listed as being decommissioned."
      }
    },
    {
      "id": "free-program-8cbd9033dece22fac992",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-10-07) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint, which is not charged and remains usable even when the account balance is zero. The free lane is best-effort with no speed or availability guarantee. An exhausted limit returns HTTP 429, and Kenari says to wait out the Retry-After period when the response carries that header.",
        "models_detail": "Fifteen provider-hosted :free text models per the Solo card on Kenari's homepage on 2026-10-07 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (13 models on 2026-10-08) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee, and once the rate limit is reached Kenari returns HTTP 429 and advises waiting for the Retry-After header when the response carries one.",
        "models_detail": "Thirteen provider-hosted :free text models per the Solo card on Kenari's homepage on 2026-10-08 (the same card read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog). Kenari's quickstart notes that the free-model list changes over time and points to a public models endpoint for the list that is active now."
      }
    },
    {
      "id": "free-program-8334c5be29f51b73fc20",
      "program_id": "together-ternary-bonsai-free",
      "event_type": "program_terms_changed",
      "program_name": "Prism-ML/Ternary-Bonsai-27B",
      "provider_label": "Together",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://www.together.ai/pricing",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "Free model after a minimum $5 upfront credit purchase; a positive balance is required. Ternary Bonsai 27B lists $0 input and $0 output token prices, so those tokens do not consume the purchased balance under published pricing. Other paid usage can drain the balance; API access stops at zero. Model rate limits are dynamic, not a guaranteed unlimited allowance.",
        "card": "A payment method and minimum $5 purchase are required. A card is not strictly the only documented payment route; Together also documents US-bank-account ACH, while card or debit card is required for auto-recharge."
      },
      "after": {
        "free_terms": "Free model behind a one-time minimum credit purchase. Together AI does not currently offer free trials, and platform access requires a minimum $15 credit purchase for an organization's first purchase, then a $5 minimum for later purchases. The platform is fully prepaid, so a positive credit balance is required and API access is suspended if the balance reaches zero. Ternary Bonsai 27B lists $0.00 input and $0.00 output token prices, so those tokens do not consume the purchased balance under published pricing. Other paid usage can drain the balance. Serverless rate limiting can occur with high request volumes or large bursts of traffic and serverless performance is best effort, so this is not a guaranteed unlimited allowance.",
        "card": "A payment method and a minimum credit purchase are required: $15 for an organization's first purchase and $5 for purchases after that. A card is not the only documented payment route, because Together also documents US bank account ACH purchases, which deposit credits only after the payment clears. A credit or debit card is required for auto-recharge."
      }
    },
    {
      "id": "free-program-545568075d82d8b76acf",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-08T00:00:00+00:00",
      "date": "2026-10-08",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "clock"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of any amount from $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all resetting daily. Some models carry lower caps because one call there counts as several requests: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute on gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page (read 2026-10-07), nearly all returning text, with two returning images (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "clock": "The 10 trial calls per new account never expire. After a one-time top-up of at least $1 the daily free quotas are permanent and reset every day, with no published end date."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all reset daily. Models that carry a higher request weight have lower caps, ranging from 50 requests per day and 5 per minute down to 10 per day and 1 per minute.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page, which was updated on 2026-10-07. Nearly all return text, and two return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free). Every free model answers the Chat Completions, Messages, and Responses protocols. The roster changes over time.",
        "clock": "Ten trial calls that never expire per new account, then permanent daily quotas after a one-time $1 top-up, reset every day with no published end date."
      }
    },
    {
      "id": "free-program-fc538ba52217a7ff5794",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "On 2026-10-01 the models whose pricing tables mark Standard Free tier input and output as free of charge are Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries, Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first-generation Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview are no longer listed on the pricing page. This is model-specific: other listed models, including the Pro Preview, image, video, and music models, show Free tier 'Not available', and some batch, flex, and priority modes do the same.",
        "clock": "No end date is stated for the Free tier. Requests per day quotas reset at midnight Pacific time, and other limits are enforced per minute or per day depending on the model."
      },
      "after": {
        "models_detail": "On 2026-10-07 the Gemini Developer API pricing page marks standard Free tier input and output as free of charge for Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. The pricing page does not list Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini Robotics ER 1.6 Preview, or Gemini 2.5 Flash-Lite Preview; those names appear only in the Batch enqueued tokens table on the rate limits page, which covers paid tiers. Free access is specific to both the model and the serving mode: Gemini 3.1 Pro Preview, Gemini Omni Flash, the Nano Banana image models, Veo 3.1, and the Lyria music models show Free tier 'Not available,' and the batch, flex, and priority tables are free of charge for some models and not available for others.",
        "clock": "Google states no end date for the Free tier. Requests per day quotas reset at midnight Pacific time, and Google states that rate limits are subject to change."
      }
    },
    {
      "id": "free-program-f4320b338362cf5d3703",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page, which presents itself as showing real-time prices and available models only, usable within platform rate limits. Rate limits for free models are fixed values, and free model calls appear in the account bill at a cost of 0. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free on that date: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. On 2026-10-06 the pricing page adds a free chat model from ChinaTelecom, XingChenAGI/Xing4.0-29B, priced free for both input and output; the model center shows it at zero cost per million tokens with a 256K context and a 29B parameter size, released on 2026-09-17. On 2026-10-06 the pricing page still shows free rows for tencent/Hunyuan-MT-7B, PaddlePaddle/PaddleOCR-VL-1.5, the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 models, Kwai-Kolors/Kolors for images, and several speech models including Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall and the ChinaTelecom XingChen ASR and GSR models. The Z-ai, deepseek-ai and Qwen vendor blocks keep 2, 6 and 30 further models behind expand buttons, so the rows visible without expanding are not the full roster. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Real-name identity verification on the account is also required: the platform's rate limits documentation states that all free models become usable after identity is verified."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page, which presents itself as syncing prices in real time and showing only available models, usable at no cost within platform rate limits. Rate limits for the free models are fixed values rather than tiered by spending, and free model calls show as a cost of 0 in the account bill. Specific per-model rate limits are published on the model pages. The roster is dynamically maintained and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-10-07, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and XingChenAGI/Xing4.0-29B from ChinaTelecom, which the model centre lists with a release date of 2026-09-17. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn with login by SMS or email, and an API key created in the console. Identity verification is also required: the platform states that after verifying your identity you can use all free models."
      }
    },
    {
      "id": "free-program-ea2b2bff33562c2db0f5",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-06) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models on 2026-10-06, per the Solo plan card on Kenari's homepage, which reads 50 requests per day for 16 free models. Earlier reads of the same homepage recorded seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-10-07) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint, which is not charged and remains usable even when the account balance is zero. The free lane is best-effort with no speed or availability guarantee. An exhausted limit returns HTTP 429, and Kenari says to wait out the Retry-After period when the response carries that header.",
        "models_detail": "Fifteen provider-hosted :free text models per the Solo card on Kenari's homepage on 2026-10-07 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      }
    },
    {
      "id": "free-program-bece44661ddd36c42d55",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "MythoLite is listed with a token price of FREE for both input and output on Mancer's model table as of 2026-10-05. Mancer states that some models are free while most carry a per token credit cost: MythoMax, for example, was priced at 0.14 credits per input token and 0.24 credits per output token on 2026-10-05. Using the free models requires no credit purchase. Mancer presents them as a way to test the service before spending money, and the prices page states that free models remain usable even when an account's credit balance is negative.",
        "models_detail": "MythoLite is the only model listed at a FREE token price, a LLaMA 2 13B model tagged for creative and roleplay use, shown as online on 2026-10-05. Its context window and maximum output length are not stated.",
        "signup": "required: Mancer is used through an account with its own API key, and the prices page notes that the usage dashboard and billing pages require login"
      },
      "after": {
        "free_terms": "MythoLite is priced FREE for both input and output tokens on Mancer's model table, so it can be used without buying credits. Mancer's pricing page says it offers free models to test with before making a purchase, and that free models can still be used when an account's credit balance is negative. The same page states that some models are free while most have a per token credit cost, so the free roster is not necessarily limited to MythoLite. Mancer provides no interface of its own, so the free models are reached through its API from another application that supports it.",
        "models_detail": "MythoLite, a LLaMA 2 13B model tagged Creative and Roleplay, carried a FREE price for both input and output tokens on Mancer's model table on 2026-10-07. The table does not state its context length or its maximum output. The other LLaMA 2 13B model listed, MythoMax, was priced at 0.14 credits per input token and 0.24 credits per output token on the same date.",
        "signup": "required: Mancer's terms of use direct users to provide accurate information on the platform's registration form and forbid sharing a login, password, or API key, and the pricing page refers to the email used to create the account."
      }
    },
    {
      "id": "free-program-acce4dde6f1c7bbc2929",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 13 free models on 2026-10-06: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Ling 3.1 Flash, Ling 3.0 Tiny, Nemotron 3.5 Lightning 30B A3B and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training, and that training and retention policies vary by the underlying model provider, with each model detail page showing that provider's policy. It describes most major providers, naming Anthropic, Vertex AI, Azure OpenAI and AWS Bedrock, as not using API data for training by default. Check Requesty and the underlying provider policies before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free models page listed 13 free models on 2026-10-07: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. Each is priced at Free for input and output. The documentation page carries a shorter nine row table that does not include Ling 3.0 Tiny, Ling 3.1 Flash, Muse Glimmer 30B or Nemotron 3.5 Lightning. Requesty says the models are free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training, and that most major providers do not use API data for training by default, with each model detail page carrying that provider's retention and training policy. Because free routes run through different model providers, treatment is not stated uniformly; check Requesty and the underlying provider policies before sensitive use."
      }
    },
    {
      "id": "free-program-92a0e4e793fa5a2b4fc0",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 4 (LLM) and Document Parse, as named on the program page on 2026-10-02. The API pricing page lists the older Solar Pro 2 and Solar Pro 3 with an end of service date of 2026-10-30 KST.",
        "signup": "Application and approval required; eligible K-12 and university institutions, university hospitals, nonprofits, and NGOs. Applicants must create an Upstage Console account using the same email address before submitting the application form. Approved organizations receive setup instructions by email, and API access follows approval.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved access lasts up to one year, and the program page states that access runs from approval until 2027-03-31, so the one year is capped by that date."
      },
      "after": {
        "models_detail": "On 2026-10-07 the program page names Solar Pro 4 (LLM) and Document Parse as the models included free for approved organizations. Solar Pro 2 and Solar Pro 3 are no longer named in the program, and the API pricing page marks both for end of service on 2026-10-30 (KST).",
        "signup": "Application and approval required. Eligible applicants are K-12 schools and universities, university hospitals, and nonprofit and non-governmental organizations. Applicants must sign up for the Upstage Console with the same email address before submitting the application form, and approved organizations receive detailed instructions by email.",
        "clock": "Applications are accepted through March 31, 2027 at 11:00 PM KST. Approved organizations get free access for up to one year, running from the date of approval until March 31, 2027, so an approval granted late in the window gives less than a full year."
      }
    },
    {
      "id": "free-program-46eb14a1c8ffcd655e63",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible usage tiers 3-5 receive up to 1,000,000 tokens/day in the large-model group and 10,000,000 tokens/day in the small-model group; tiers 1-2 receive 250,000 and 2,500,000 tokens/day respectively. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed normally. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded.",
        "models_detail": "As of 2026-10-06 the 1,000,000 and 250,000 token group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10,000,000 and 2,500,000 token group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the larger group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "free_terms": "Explicitly limited free promotion for eligible API organizations that opt in to share API inputs and outputs. Eligible Launch and Grow organizations receive up to 1,000,000 tokens per day in the large-model group and 10,000,000 tokens per day in the small-model group. Eligible Build organizations receive up to 250,000 and 2,500,000 tokens per day respectively. Quotas are shared within each group and reset at 00:00 UTC. A request that would cross the daily quota is billed in full, and overage is billed at normal rates. Only traffic from sharing-enabled projects qualifies. A positive account balance is required. Fine-tuned models, fine-tuning training, evals, and tool use are excluded. Eligibility is shown on the data sharing settings page, and an organization that does not see the offer there is not eligible.",
        "models_detail": "On 2026-10-07 the 1M/250K group was: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group was: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The larger quota in each pair applies to Launch and Grow organizations and the smaller to Build organizations. The page also retains gpt-4.5-preview-2025-02-27 in the first group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-397d0d3c4ae998e79ee7",
      "program_id": "catalog2-regolo-ai",
      "event_type": "program_terms_changed",
      "program_name": "Regolo brick-v1-beta at zero price",
      "provider_label": "Regolo AI",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://regolo.ai/pricing/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock",
        "ends_at",
        "access_clock"
      ],
      "before": {
        "free_terms": "Regolo's pricing page and model catalog list brick-v1-beta at 0.00 EUR input and 0.00 EUR output under pay-as-you-go, and the PAYG plan has a 0 EUR monthly fee, so this is an ongoing zero-token-price hosted route distinct from Regolo's separate 30-day trial. Regolo's deepseek-ocr-2 and faster-whisper-large-v3 are billed per request and per audio second, not per token, so only brick-v1-beta qualifies.",
        "models_detail": "brick-v1-beta only. deepseek-ocr-2 and faster-whisper-large-v3 use request/audio pricing and are not free inference routes.",
        "data_treatment": "Regolo presents the service on its pricing page under the heading Zero Data Retention, Fast and Secure AI, and the plan copy on that page describes secure, zero-retention AI. The pricing, models library, model catalog, and sign-up pages do not state whether prompts or outputs are used for model training, so review Regolo's privacy policy and DPA before sensitive use.",
        "clock": "Ongoing beta model with no published end date, and beta pricing and availability may change. Regolo's pricing page carries a notice that it has updated its rates and that the new rates go live on November 1, and it states that current pricing rates and plan rules remain active until that date, so the zero price on brick-v1-beta is not guaranteed beyond that change.",
        "ends_at": null,
        "access_clock": "standing"
      },
      "after": {
        "free_terms": "Regolo's pricing page lists brick-v1-beta at \u20ac0.00 input and \u20ac0.00 output per 1M tokens under pay-as-you-go, and the models library shows the same model's input and output as Free, so a pay-as-you-go account can call it at a zero token price. Pay-as-you-go carries a \u20ac0 monthly fee and bills only for the tokens consumed, which makes this route distinct from Regolo's separate 30-day free trial. Every other model in the library carries a paid per-token rate or is billed per request, per second of audio, per pixel, or per query, so brick-v1-beta is the only zero-cost inference route here. On 2026-10-07 the pricing page also carried a notice that updated rates go live on November 1 and that current pricing rates and plan rules remain active until that date, and the reviewed pages do not state what brick-v1-beta will cost after that change.",
        "models_detail": "On 2026-10-07 brick-v1-beta was the only model listed at a zero token price. apertus-70b, brick-complexity-pro, gemma4-31b, glm5.2, gpt-oss-120b, gpt-oss-20b, mistral-small-4-119b, qwen3.5-122b, qwen3.5-9b, and qwen3.8-27b all carried paid per-token rates. deepseek-ocr-2, faster-whisper-large-v3, Qwen-Image, Qwen3-Embedding-8B, and Qwen3-Reranker-4B are billed per request, per second of audio, per pixel, or per query rather than per token, so none of them is a free inference route.",
        "data_treatment": "Regolo presents the service on its pricing page under the heading Zero Data Retention, Fast and Secure AI, and links separate terms, privacy policy, and DPA documents. The reviewed pricing, models, and documentation pages do not state whether prompts sent to brick-v1-beta are used for training, so read Regolo's privacy terms and DPA before sensitive use.",
        "clock": "Regolo lists brick-v1-beta as a beta model and publishes no retirement date for the model itself. On 2026-10-07 the pricing page carried a notice that updated rates go live on November 1, 2026, and that current pricing rates and plan rules remain active until that date, so the \u20ac0.00 listing is only stated to hold until then. The reviewed pages do not say what brick-v1-beta will cost after that change, and beta pricing and availability may change.",
        "ends_at": "2026-11-01",
        "access_clock": "limited_dated"
      }
    },
    {
      "id": "free-program-2f4b5c4db3a2c1efc0fe",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Thirteen free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen prices twelve models at Free on its pricing table as of 2026-10-06. All twelve are Free for input and output tokens, eleven of them are also Free for cached reads, and Jev 1.13 Free lists no cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-06: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model used through Zen's systemone endpoint rather than a chat endpoint. Union Alpha Free is no longer on the free table as of 2026-10-06. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy, and the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen says its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: Zen says not to submit personal or confidential data, and use is logged for security purposes and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use data for training. For Jev, prompts and other inputs are not used for training.",
        "clock": "All twelve free models are described as available for a limited time, and no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen prices thirteen models at Free on its pricing table as of 2026-10-07, free for input and output tokens, and free for cached reads on every one of them except Jev 1.13 Free, which lists no cached read price. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-07: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model called through the systemone endpoint rather than a text generation model. Union Alpha Free left the free table by 2026-10-07, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, Exo Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free may use collected data to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints whose use is logged for security and to improve NVIDIA products and services, and users are told not to submit personal or confidential data. Muse Spark 1.3 Contributor Free carries permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use the data for training, and prompts sent to the Jev endpoints behind Jev 1.13 Free are not used for training.",
        "clock": "All thirteen free models are described as limited-time promotions; no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-1a23902bffb2266a79a8",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free models and reset every day. Models that carry a higher request weight have lower request caps, as low as 10 per day and 1 per minute, and past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "On 2026-10-06 the provider's free catalog page listed 60 models at $0 input and $0 output, nearly all of them text models plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The page carries its own roster stamp of 2026-10-05, and the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Reaching the daily free quotas requires a one-time top-up of at least $1, which is a payment rather than a card kept on file, and usage within the quotas afterwards bills $0 and does not draw that balance down.",
        "card_required": "yes"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of any amount from $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all resetting daily. Some models carry lower caps because one call there counts as several requests: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute on gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page (read 2026-10-07), nearly all returning text, with two returning images (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the ten trial calls. Reaching the recurring daily quotas takes a one-time top-up of any amount from $1, and the provider does not state which payment methods it accepts for that top-up.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-170b1b307d42bdb3e385",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-05 the catalog listed 20 models, drawn from the Llama, Qwen, Qwen3Guard, GPT-OSS, Whisper, BGE, NVIDIA Riva, and Stable Diffusion families, and covering text, reasoning, visual, embedding, image-generation, and speech models. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalog also prices seven entries at Free outright on 2026-10-05: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices nvr-tts-de-de, nvr-tts-en-us, nvr-tts-es-es, and nvr-tts-it-it. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR per 1M input tokens and 2.7 EUR per 1M output tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. OVHcloud describes the offer as more than 20 open-weight models, and the catalogue listed 20 models on 2026-10-07, covering Llama, Qwen, Qwen3Guard, GPT-OSS, bge embeddings, Stable Diffusion image generation, Whisper speech to text, and NVIDIA Riva text to speech. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalogue also prices a few entries at Free outright, on 2026-10-07 these were the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text to speech voices, while Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      }
    },
    {
      "id": "free-program-100c4cd9dc796fea8cba",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Each account gets a free allowance of 10 requests per minute and 60 weighted units per day while it holds no credit, rising to 20 requests per minute and 120 weighted units per day once the account has credit. Rate limits use a rolling 60 second window and the daily counter resets at 00:00 UTC. The allowance is counted in weighted units rather than plain request counts, so a long prompt consumes more units and long-context work exhausts the day faster. A site-wide free cap of 15 requests per minute and a limit of 2 concurrent free requests per account also apply, alongside an additional IP guard. Past the allowance, requests continue at the paid rate only when the account holds credit and paid fallback is enabled; sending the X-Free-Fallback: false header turns that fallback off, and an account with no balance gets a rate-limit response until the allowance resets. BazaarLink's free tier page says the allowance is shared across the free models, while the API docs say each free model is counted separately. The 200 requests per minute figure in the docs is the paid-tier account limit, not the free allowance.",
        "models_detail": "Two free-quota models on 2026-10-06: Qwen3.7 Flash (qwen/qwen3.7-flash, 1,000,000 token context, text, image and video input) and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free, 1,048,576 token context, text input). The auto:free route always resolves to a free model. The catalog changes over time."
      },
      "after": {
        "free_terms": "Accounts with no credit get 10 requests per minute and 60 weighted units per day. Accounts that have topped up get 20 requests per minute and 120 weighted units per day, so a top-up doubles both the per-minute limit and the daily allowance. On 2026-10-07 the free page states that this allowance is shared across the free models for an account, with an additional IP guard, a site-wide free cap of 15 requests per minute, and a limit of 2 concurrent free requests per account. The per-minute limit uses a rolling 60 second window and the daily counter resets at 00:00 UTC. When weighted limits are enabled, longer prompts use more daily units, so long-context requests use the allowance up faster; otherwise each request counts once. The developer docs give the general rule as a separate daily count for each free model and add that a model's own listed allowance governs, which is the shared figure shown on the free page. Past the allowance, requests continue at the paid rate only when the account holds credit, and sending the header X-Free-Fallback: false turns that fallback off; an account with no balance gets 429 until the quota resets. The docs' example rate-limit header (200 requests per minute) is the paid-tier account limit, not the free quota.",
        "models_detail": "Two enabled free-quota models on 2026-10-07: Qwen3.7 Flash (qwen/qwen3.7-flash), with a 1,000,000 token context and text, image and video input, and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), with a 1,048,576 token context and text input. The auto:free route always resolves to a free model. The catalogue changes over time, and only enabled free models are listed."
      }
    },
    {
      "id": "free-program-0ec5d5eb29d693ede36f",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) contain no billing or fee language. The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and that the alternative AWS IAM authentication instead bills usage to an AWS account. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). No numeric daily limit is published publicly."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), and an API key is tied to the holder's Amazon.com account. As documented on 2026-10-07, the AWS Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication method instead applies usage charges to an AWS account. The Terms add that usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the offering does not support deploying production scale Developer Offerings (S-1.1). Amazon publishes no numeric daily limit, quota, or rate limit for the free route."
      }
    },
    {
      "id": "free-program-03b45c2b22c3d69ee717",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-07T00:00:00+00:00",
      "date": "2026-10-07",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-06: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also paid, at $0.13 in and $0.53 out per 1M. Some other catalog entries show a $0.00 input price alongside a paid output price, which is not free inference.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-07: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. Several of the free models appear a second time in the catalog as separate paid entries at list prices.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms state that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-event-00924af628c75704e773",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "qwen/qwen3.8-27b:free",
      "model": "Qwen: Qwen3.8 27B",
      "model_key": "qwen3827bfree",
      "detected_at": "2026-10-06T11:30:18+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://openrouter.ai/qwen/qwen3.8-27b:free"
    },
    {
      "id": "free-program-e44a6698722b051329de",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states, on 2026-10-05, that API key authentication is free of charge with daily limits, using a key from nova.amazon.com tied to an Amazon.com account, and that the separate AWS IAM authentication route instead bills usage charges to an AWS account. No numeric limits are published publicly.",
        "card": "Not required. Amazon states that API key authentication is free of charge with daily limits, and the Nova.amazon.com Terms of Use contain no billing or fee language. Charges arise only on the optional AWS IAM authentication route, which bills an AWS account."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) contain no billing or fee language. The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and that the alternative AWS IAM authentication instead bills usage to an AWS account. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). No numeric daily limit is published publicly.",
        "card": "not required; the Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and the Nova Terms of Use contain no payment terms. The paid alternative is AWS IAM authentication, which bills usage to an AWS account."
      }
    },
    {
      "id": "free-program-bf74247ff44587bb04c8",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Twelve free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table. Nine of them are Free for input, output, and cached reads, and Jev 1.13 Free is Free for input and output with no cached read price listed. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-02: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's Jev System One endpoint, which evaluates typed questions instead of generating text, and the two Nemotron entries are NVIDIA free trial endpoints. Union Alpha Free no longer appears on the free table on 2026-10-02, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the provider tells users not to submit personal or confidential data. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. The providers behind Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and do not use data for model training. Jev 1.13 Free runs on the Jev APIs, where prompts and other inputs are not used for training and data is retained under TypeSafe AI's privacy policy.",
        "clock": "All ten are limited-time promotions; no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen prices twelve models at Free on its pricing table as of 2026-10-06. All twelve are Free for input and output tokens, eleven of them are also Free for cached reads, and Jev 1.13 Free lists no cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-06: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model used through Zen's systemone endpoint rather than a chat endpoint. Union Alpha Free is no longer on the free table as of 2026-10-06. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy, and the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen says its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, Fledge Alpha Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.1 Flash Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: Zen says not to submit personal or confidential data, and use is logged for security purposes and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free come from providers that follow a zero-retention policy and do not use data for training. For Jev, prompts and other inputs are not used for training.",
        "clock": "All twelve free models are described as available for a limited time, and no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-b4484c5e5d94617705c5",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-05) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier listing on Kenari's homepage on 2026-10-05 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-06) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models on 2026-10-06, per the Solo plan card on Kenari's homepage, which reads 50 requests per day for 16 free models. Earlier reads of the same homepage recorded seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog."
      }
    },
    {
      "id": "free-program-851a47b5eb99d6aca8f8",
      "program_id": "groq-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Groq free-tier models",
      "provider_label": "Groq",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://console.groq.com/docs/rate-limits.md",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-10-05: openai/gpt-oss-120b and openai/gpt-oss-20b 30 RPM, 1K RPD, 8K TPM, 200K TPD; openai/gpt-oss-safeguard-20b 5 RPM, 1K RPD, 2K TPM, 200K TPD; qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and meta-llama/llama-prompt-guard-2-86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and canopylabs/orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio-seconds per hour, 28.8K audio-seconds per day. Rate limits apply at the organization level rather than per user, and cached tokens do not count toward them. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table on 2026-10-05: openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-09-02: groq/compound, groq/compound-mini, and qwen/qwen3.6-27b. Removed earlier, since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      },
      "after": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-10-06: openai/gpt-oss-120b, openai/gpt-oss-20b, and qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; openai/gpt-oss-safeguard-20b 3 RPM, 1K RPD, 2K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and meta-llama/llama-prompt-guard-2-86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and canopylabs/orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio seconds per hour, 28.8K audio seconds per day. Rate limits apply at the organization level rather than to individual users, and cached tokens do not count toward them. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table as of 2026-10-06: openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-09-01: groq/compound, groq/compound-mini, and qwen/qwen3.6-27b. Removed from the table earlier, since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      }
    },
    {
      "id": "free-program-6e4e02dd3a31afcf254e",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-05: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 remains outside the free set, at $0.13 in and $0.53 out per 1M. Several of these model names also appear as separate paid entries at normal rates, so the free route is the entry shown at $0.00/1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms say that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-06: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Mini 4. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also paid, at $0.13 in and $0.53 out per 1M. Some other catalog entries show a $0.00 input price alongside a paid output price, which is not free inference.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-63f674bc7f155ac8de7b",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-30, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and XingChenAGI/Xing4.0-29B from China Telecom. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is listed.",
        "signup": "required; account at cloud.siliconflow.cn, with the platform supporting login by SMS and by email, and the API key created in the console. The platform documentation states that after verifying your identity you can use all free models, so identity verification is part of reaching the free roster."
      },
      "after": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free on that date: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. On 2026-10-06 the pricing page adds a free chat model from ChinaTelecom, XingChenAGI/Xing4.0-29B, priced free for both input and output; the model center shows it at zero cost per million tokens with a 256K context and a 29B parameter size, released on 2026-09-17. On 2026-10-06 the pricing page still shows free rows for tencent/Hunyuan-MT-7B, PaddlePaddle/PaddleOCR-VL-1.5, the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 models, Kwai-Kolors/Kolors for images, and several speech models including Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall and the ChinaTelecom XingChen ASR and GSR models. The Z-ai, deepseek-ai and Qwen vendor blocks keep 2, 6 and 30 further models behind expand buttons, so the rows visible without expanding are not the full roster. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Real-name identity verification on the account is also required: the platform's rate limits documentation states that all free models become usable after identity is verified."
      }
    },
    {
      "id": "free-program-62eb52075ce3492ec475",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms",
        "clock"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan.",
        "clock": "Monthly allowance with quota limits that reset on the first of every month. No fixed offer end date is stated."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, and 100 text-extraction pages per month, which IBM's pricing page states as 100 documents per month. The plan carries a limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan.",
        "clock": "Monthly allowance: IBM states that Lite plan quota limits reset on the first of every month, and that reaching a quota limit suspends the service for the rest of that month. No fixed offer end date is stated."
      }
    },
    {
      "id": "free-program-50a9cb913567687caaa5",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-10-05 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M list but marks it deprecated and shut down as of 7/14/25, so it is not current access. The quota is shared across each group, and fine-tuned models, fine-tuning training, evals, and tool use are not included."
      },
      "after": {
        "models_detail": "As of 2026-10-06 the 1,000,000 and 250,000 token group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10,000,000 and 2,500,000 token group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the larger group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-4bc2cf07277396418256",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-10-02: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The Requesty documentation page carries a shorter table of nine of these models, leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policies for the underlying model providers vary, and each model's detail page shows the policy that applies to that provider, so check them before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 13 free models on 2026-10-06: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny and Ling 3.1 Flash; and Meta Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Ling 3.1 Flash, Ling 3.0 Tiny, Nemotron 3.5 Lightning 30B A3B and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training, and that training and retention policies vary by the underlying model provider, with each model detail page showing that provider's policy. It describes most major providers, naming Anthropic, Vertex AI, Azure OpenAI and AWS Bedrock, as not using API data for training by default. Check Requesty and the underlying provider policies before sensitive use."
      }
    },
    {
      "id": "free-program-1cd9ff4a45db4f543415",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "Free use is capped per account and, on the free tier page, shared across the current free-quota models: accounts with no credit get 10 requests per minute and 60 weighted daily units, and accounts that hold credit get 20 requests per minute and 120 weighted daily units. The rate limit uses a rolling 60 second window and the daily counter resets at 00:00 UTC. The allowance is counted in weighted units rather than plain requests, so longer inputs consume more units and long context requests use the daily allowance up faster. A site wide free cap of 15 requests per minute, a limit of 2 concurrent free requests per account, and an additional IP guard also apply. The two provider surfaces differ on one point: the free tier page says the daily allowance is shared across free models, while the docs page says it is counted separately for each free model. Past the allowance, requests continue at the paid rate only when paid fallback is enabled and the account holds credit, and sending the X-Free-Fallback: false header returns 429 instead; an account with no balance gets 429 until the allowance resets. The account rate limit tiers in the docs, 20 requests per minute below $5 of credit and 200 requests per minute at or above it, are account limits rather than the free quota.",
        "models_detail": "Two free-quota models were enabled on 2026-10-04: Qwen3.7 Flash (qwen/qwen3.7-flash), with a 1,000,000 token context and text, image and video input, and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), with a 1,048,576 token context and text input. The auto:free route always resolves to a free model. Only enabled free models are listed, and the catalogue changes over time.",
        "data_treatment": "Provider says it does not store message content by default, processing it in memory and discarding it immediately, and that it keeps billing metadata such as token counts, timestamps and model IDs. Every request passes automated content moderation, and the provider says it retains the content of requests judged to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to those providers' privacy policies."
      },
      "after": {
        "free_terms": "Each account gets a free allowance of 10 requests per minute and 60 weighted units per day while it holds no credit, rising to 20 requests per minute and 120 weighted units per day once the account has credit. Rate limits use a rolling 60 second window and the daily counter resets at 00:00 UTC. The allowance is counted in weighted units rather than plain request counts, so a long prompt consumes more units and long-context work exhausts the day faster. A site-wide free cap of 15 requests per minute and a limit of 2 concurrent free requests per account also apply, alongside an additional IP guard. Past the allowance, requests continue at the paid rate only when the account holds credit and paid fallback is enabled; sending the X-Free-Fallback: false header turns that fallback off, and an account with no balance gets a rate-limit response until the allowance resets. BazaarLink's free tier page says the allowance is shared across the free models, while the API docs say each free model is counted separately. The 200 requests per minute figure in the docs is the paid-tier account limit, not the free allowance.",
        "models_detail": "Two free-quota models on 2026-10-06: Qwen3.7 Flash (qwen/qwen3.7-flash, 1,000,000 token context, text, image and video input) and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free, 1,048,576 token context, text input). The auto:free route always resolves to a free model. The catalog changes over time.",
        "data_treatment": "Provider says it does not store message content by default, processing it in memory and keeping only billing metadata such as token counts, timestamps and model IDs. Every request passes automated content moderation, and the provider says it retains the content of requests judged to violate its policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to their privacy policies."
      }
    },
    {
      "id": "free-program-1baae4f276efb21b6e24",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "clock"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all resetting daily. Models that carry a higher request weight have lower caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute, and the per model caps are listed in the provider's free catalog table.",
        "models_detail": "60 models listed at $0 for input and output on the provider's free catalog page, which was marked updated on 2026-10-04 (checked on 2026-10-05). Nearly all return text, two return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free), and many accept vision, audio, video, or PDF input. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, which never expire. Reaching the permanent daily quotas requires a one-time top-up of $1 or more, and free model usage still bills $0 for input and output tokens.",
        "clock": "The ten trial calls given to each new account never expire. After a one-time top-up of $1 or more the daily quotas are permanent and reset every day, and the provider publishes no end date."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free models and reset every day. Models that carry a higher request weight have lower request caps, as low as 10 per day and 1 per minute, and past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "On 2026-10-06 the provider's free catalog page listed 60 models at $0 input and $0 output, nearly all of them text models plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The page carries its own roster stamp of 2026-10-05, and the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Reaching the daily free quotas requires a one-time top-up of at least $1, which is a payment rather than a card kept on file, and usage within the quotas afterwards bills $0 and does not draw that balance down.",
        "clock": "The 10 trial calls per new account never expire. After a one-time top-up of at least $1 the daily free quotas are permanent and reset every day, with no published end date."
      }
    },
    {
      "id": "free-program-17ecc8b8cedbc4ab5ced",
      "program_id": "catalog2-regolo-ai",
      "event_type": "program_terms_changed",
      "program_name": "Regolo brick-v1-beta at zero price",
      "provider_label": "Regolo AI",
      "detected_at": "2026-10-06T00:00:00+00:00",
      "date": "2026-10-06",
      "evidence_url": "https://regolo.ai/pricing/",
      "changed_fields": [
        "data_treatment",
        "clock"
      ],
      "before": {
        "data_treatment": "Regolo's pricing page presents the service under a Zero Data Retention banner and describes its plans as zero-retention AI. The reviewed pricing, model, and catalog pages do not say whether prompts or outputs are used for training, so check the provider privacy terms before sensitive use.",
        "clock": "Ongoing beta model with no published end date; beta pricing and availability may change."
      },
      "after": {
        "data_treatment": "Regolo presents the service on its pricing page under the heading Zero Data Retention, Fast and Secure AI, and the plan copy on that page describes secure, zero-retention AI. The pricing, models library, model catalog, and sign-up pages do not state whether prompts or outputs are used for model training, so review Regolo's privacy policy and DPA before sensitive use.",
        "clock": "Ongoing beta model with no published end date, and beta pricing and availability may change. Regolo's pricing page carries a notice that it has updated its rates and that the new rates go live on November 1, and it states that current pricing rates and plan rules remain active until that date, so the zero price on brick-v1-beta is not guaranteed beyond that change."
      }
    },
    {
      "id": "free-program-f7954eb88c69feba77c7",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_record_corrected",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "data_treatment"
      ],
      "before": {
        "data_treatment": "Hetzner's Inference API documentation states that the service keeps only what it needs for usage accounting: \"We do only keep data that is necessary in order to track usage and (in the future) bill usage: request timestamps, token counts etc.\" The same page states, \"We do not store the content of request and response and we do not plan on doing so in the future (unless some law requires us to do so).\" Hetzner's engineering blog adds, \"We do not log any prompt data.\""
      },
      "after": {
        "data_treatment": "The Inference API FAQ states: \"We do only keep data that is necessary in order to track usage and (in the future) bill usage: request timestamps, token counts etc.\" It also states: \"We do not store the content of request and response and we do not plan on doing so in the future (unless some law requires us to do so).\" Hetzner's blog post on the experiment adds, \"We do not log any prompt data.\" No page states anything about using customer data for model training."
      }
    },
    {
      "id": "free-program-cab62991b56dcbe426dd",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-10-04 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in its list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "On 2026-10-05 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M list but marks it deprecated and shut down as of 7/14/25, so it is not current access. The quota is shared across each group, and fine-tuned models, fine-tuning training, evals, and tool use are not included."
      }
    },
    {
      "id": "free-program-b189e26371cd3ae7f3c7",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. Access runs on an Amazon-provided API key (Terms 1.5) issued from the developer surface, and the Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, with the key tied to an Amazon.com account. Usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production scale developer applications (S-1.1). No numeric limits are published publicly. A separate AWS IAM authentication route exists for Nova Act and is not free: usage charges apply to the AWS account, so the zero-cost route is the API key one only."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states, on 2026-10-05, that API key authentication is free of charge with daily limits, using a key from nova.amazon.com tied to an Amazon.com account, and that the separate AWS IAM authentication route instead bills usage charges to an AWS account. No numeric limits are published publicly."
      }
    },
    {
      "id": "free-program-6911a5e410bf226e9157",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-10-04: inclusionAI Ling 3.0 Flash Fin, inclusionAI Ling 3.0 Flash Sante (free), inclusionAI Ling 3.1 Flash, Meituan LongCat 2.0, Meituan LongCat 2.5 Preview, Poolside Laguna S 2.1, Poolside Laguna XS 2.1, Space Bunny Alpha, and StepFun Step 3.7 Flash. Several of those names also appear as separate paid entries in the same catalog, so the $0.00/1M listing is the free one. Upstage Solar Pro 4 has left the free set and is now listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. The catalog is advertised at 300 or more models overall, and which entries are marked free can change.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it, and allow Nous to make Service Data and derived models available to third parties. The same terms state that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The terms do not say whether Privacy Mode is available on the $0 Free plan, and the plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Nine catalog entries were priced at $0.00/1M on 2026-10-05: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); inclusionAI Ling 3.1 Flash; Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 remains outside the free set, at $0.13 in and $0.53 out per 1M. Several of these model names also appear as separate paid entries at normal rates, so the free route is the entry shown at $0.00/1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms say that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-604804cea0d6199c1c6a",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Mancer's model table shows MythoLite with an input and output token cost of FREE on 2026-09-29, so it can be used without buying credits. Mancer's pricing FAQ says some models are free while most have a per token credit cost, and it does not publish the full free list, so MythoLite is the only model the table shows as free. Of the other listed models, only MythoMax carries a visible price on Mancer's pages: 0.14 credits per input token and 0.24 credits per output token. Free models remain usable even when an account's credit balance is negative.",
        "models_detail": "MythoLite is a LLaMA 2 13B model tagged for creative and roleplay use, and on 2026-09-29 it is the one model on Mancer's table shown with a FREE input and output token cost. The table's context and limits columns carry no values for it, and Mancer does not state its context window or maximum output length, so those caps are not stated here.",
        "signup": "An account is required. Mancer's pricing page notes that the usage dashboard and the billing page both require login, and Mancer provides no interface of its own, so a separate client application is needed to send requests to the API."
      },
      "after": {
        "free_terms": "MythoLite is listed with a token price of FREE for both input and output on Mancer's model table as of 2026-10-05. Mancer states that some models are free while most carry a per token credit cost: MythoMax, for example, was priced at 0.14 credits per input token and 0.24 credits per output token on 2026-10-05. Using the free models requires no credit purchase. Mancer presents them as a way to test the service before spending money, and the prices page states that free models remain usable even when an account's credit balance is negative.",
        "models_detail": "MythoLite is the only model listed at a FREE token price, a LLaMA 2 13B model tagged for creative and roleplay use, shown as online on 2026-10-05. Its context window and maximum output length are not stated.",
        "signup": "required: Mancer is used through an account with its own API key, and the prices page notes that the usage dashboard and billing pages require login"
      }
    },
    {
      "id": "free-program-4a653f2659e40a359c19",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-04) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier card on Kenari's homepage on 2026-10-04 (the same counter read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-05) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier listing on Kenari's homepage on 2026-10-05 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      }
    },
    {
      "id": "free-program-37ee267ba149af89617d",
      "program_id": "groq-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Groq free-tier models",
      "provider_label": "Groq",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://console.groq.com/docs/rate-limits.md",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-09-20: openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, and qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and -86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio-seconds per hour, 28.8K audio-seconds per day. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table as of 2026-09-20: canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-09-02: groq/compound, groq/compound-mini, and qwen/qwen3.6-27b. Removed earlier, since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      },
      "after": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-10-05: openai/gpt-oss-120b and openai/gpt-oss-20b 30 RPM, 1K RPD, 8K TPM, 200K TPD; openai/gpt-oss-safeguard-20b 5 RPM, 1K RPD, 2K TPM, 200K TPD; qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and meta-llama/llama-prompt-guard-2-86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and canopylabs/orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio-seconds per hour, 28.8K audio-seconds per day. Rate limits apply at the organization level rather than per user, and cached tokens do not count toward them. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table on 2026-10-05: openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-09-02: groq/compound, groq/compound-mini, and qwen/qwen3.6-27b. Removed earlier, since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      }
    },
    {
      "id": "free-program-2a8d1b9d89087075d860",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-02 the catalog lists 20 models, covering Llama, Qwen, GPT-OSS, BGE embedding, Whisper speech-to-text, NVIDIA Riva text-to-speech, and Stable Diffusion image generation. Individual models can be retired under OVHcloud's lifecycle process, and a model announced for decommissioning stays callable through the API during its grace period while being hidden from the catalog, so the catalog count can understate what the API still serves. The authenticated catalog also prices several entries at Free outright, on 2026-10-02 the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices (Qwen3.8-27B itself is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens)."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-05 the catalog listed 20 models, drawn from the Llama, Qwen, Qwen3Guard, GPT-OSS, Whisper, BGE, NVIDIA Riva, and Stable Diffusion families, and covering text, reasoning, visual, embedding, image-generation, and speech models. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalog also prices seven entries at Free outright on 2026-10-05: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices nvr-tts-de-de, nvr-tts-en-us, nvr-tts-es-es, and nvr-tts-it-it. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR per 1M input tokens and 2.7 EUR per 1M output tokens."
      }
    },
    {
      "id": "free-program-296722b1f4c00f68fa64",
      "program_id": "catalog2-unorouter",
      "event_type": "program_record_corrected",
      "program_name": "UnoRouter free models",
      "provider_label": "UnoRouter",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://unorouter.com/en/docs/platform/models-and-pricing",
      "changed_fields": [
        "free_terms",
        "signup",
        "data_treatment",
        "clock",
        "training_use"
      ],
      "before": {
        "free_terms": "UnoRouter publishes free models that carry a :free suffix, and a :free model routes only to free providers and never uses an account balance. Calling them requires a UnoRouter account and an API key. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, and 503 when every provider for a free model is busy. UnoRouter's published pages do not state which sign-in methods are offered or whether a payment method is required to open an account. There is no announced expiry, but availability is best-effort and the free model set shifts over time.",
        "signup": "A UnoRouter account and an API key are required. UnoRouter's published pages do not state which sign-in methods are offered or whether a payment method is needed at signup.",
        "data_treatment": "Varies by routed upstream provider; check UnoRouter and selected provider policies before sensitive use.",
        "clock": "Standing best-effort catalog with a cap of 1 request per minute, per free model, per user, plus the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit. No published expiry.",
        "training_use": "varies"
      },
      "after": {
        "free_terms": "UnoRouter marks part of its catalog free: a model carrying the :free suffix routes only to free providers and never uses an account balance, while the same model name without the suffix is the paid version billed per token. Access requires a UnoRouter account and an API key. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. A rate-limited route returns 429 with a Retry-After header, and a model whose free providers are all busy returns 503, which usually clears in minutes. UnoRouter's models, pricing and rate limit pages do not state whether a credit card is required. There is no announced expiry, and availability is best effort: a model leaves the catalog only when every channel is out, and it returns within minutes once a channel recovers.",
        "signup": "A UnoRouter account and an API key are required. UnoRouter's documentation does not state which sign-in methods are available or whether a credit card is collected at signup.",
        "data_treatment": "UnoRouter's models, pricing and rate limit pages do not state how prompts sent to free models are handled or whether they can be used for training. Free requests route across upstream provider groups and can fail over between them, so check UnoRouter's privacy policy and the policy of the provider serving the model before sensitive use.",
        "clock": "Standing best-effort catalog. UnoRouter caps free models at 1 request per minute, per free model, per user, and some free pools also carry daily token budgets that reset at midnight UTC. Upstream provider caps apply on top. UnoRouter publishes no end date.",
        "training_use": "not_stated"
      }
    },
    {
      "id": "free-program-1f4a105b1cb3ee513d6c",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Accounts with no credit get 10 requests per minute and 60 weighted units per day; accounts that have topped up get 20 requests per minute and 120 weighted units per day. The free page states that this allowance is shared across the enabled free models, while the free-models documentation states that the daily quota is counted separately for each free model. Requests per minute use a rolling 60 second window and the daily counter resets at 00:00 UTC. A site-wide free cap of 15 requests per minute, a limit of 2 concurrent free requests per account, and an additional IP guard also apply. When weighted limits are enabled, longer inputs use more daily units, so long-context requests use the allowance up faster; otherwise each request counts once. Past the quota, requests continue at the paid rate only when paid fallback is enabled and the account holds credit, and sending the header X-Free-Fallback: false turns that fallback off; an account with no balance gets 429 until the quota resets. The documentation's example rate-limit header of 200 requests per minute is the paid account tier limit at $5 or more of credit, not the free quota."
      },
      "after": {
        "free_terms": "Free use is capped per account and, on the free tier page, shared across the current free-quota models: accounts with no credit get 10 requests per minute and 60 weighted daily units, and accounts that hold credit get 20 requests per minute and 120 weighted daily units. The rate limit uses a rolling 60 second window and the daily counter resets at 00:00 UTC. The allowance is counted in weighted units rather than plain requests, so longer inputs consume more units and long context requests use the daily allowance up faster. A site wide free cap of 15 requests per minute, a limit of 2 concurrent free requests per account, and an additional IP guard also apply. The two provider surfaces differ on one point: the free tier page says the daily allowance is shared across free models, while the docs page says it is counted separately for each free model. Past the allowance, requests continue at the paid rate only when paid fallback is enabled and the account holds credit, and sending the X-Free-Fallback: false header returns 429 instead; an account with no balance gets 429 until the allowance resets. The account rate limit tiers in the docs, 20 requests per minute below $5 of credit and 200 requests per minute at or above it, are account limits rather than the free quota."
      }
    },
    {
      "id": "free-program-006809aa4b76bb3782b6",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-05T00:00:00+00:00",
      "date": "2026-10-05",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls across the free catalog with no credit card and no expiry. A one-time top-up of $1 or more then switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all reset every day. Some models carry lower request caps because one call counts as several requests. On 2026-10-03 those caps were 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the Gemini image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors are listed at $0 input and $0 output on the provider's free catalog page, which was stamped Updated 2026-10-03 (read 2026-10-03). Most return text only, many also accept vision, audio, video, or PDF input, and two cover images: gpt-image-2-free returns text and images, and gemini-3.1-flash-image-preview-free returns images. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. A one-time top-up of $1 or more, which is a payment rather than a card kept on file, unlocks the daily free quotas across the free catalog."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all resetting daily. Models that carry a higher request weight have lower caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute, and the per model caps are listed in the provider's free catalog table.",
        "models_detail": "60 models listed at $0 for input and output on the provider's free catalog page, which was marked updated on 2026-10-04 (checked on 2026-10-05). Nearly all return text, two return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free), and many accept vision, audio, video, or PDF input. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, which never expire. Reaching the permanent daily quotas requires a one-time top-up of $1 or more, and free model usage still bills $0 for input and output tokens."
      }
    },
    {
      "id": "free-program-ffe2bcd1786d4ab30b56",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is free of charge with daily limits, while the separate AWS IAM authentication route instead applies usage charges to an AWS account. No numeric limits are published publicly."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. Access runs on an Amazon-provided API key (Terms 1.5) issued from the developer surface, and the Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, with the key tied to an Amazon.com account. Usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production scale developer applications (S-1.1). No numeric limits are published publicly. A separate AWS IAM authentication route exists for Nova Act and is not free: usage charges apply to the AWS account, so the zero-cost route is the API key one only."
      }
    },
    {
      "id": "free-program-ead6a1950ba866e284b3",
      "program_id": "poolside-laguna-limited-free",
      "event_type": "program_record_corrected",
      "program_name": "Laguna hosted API, free for a limited time",
      "provider_label": "Poolside",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://poolside.ai/models",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for its Laguna models, and the models page says Laguna is free to use for a limited time. On 2026-09-29 that free section covers both Laguna S 2.1 and Laguna XS 2.1, and the Laguna XS 2.1 announcement says free and paid endpoints are both available, with paid usage priced at $0.10 per 1M input tokens, $0.20 per 1M output tokens, and $0.05 per 1M cache-read tokens. No quota or rate limit for the free endpoints is published, and no end date is given.",
        "models_detail": "Hosted Laguna models are free to use for a limited time. On 2026-09-29 the free offer covers Laguna S 2.1, called as poolside/laguna-s-2.1, and Laguna XS 2.1, both served over an OpenAI-compatible chat API at https://inference.poolside.ai/v1 with the /v1/models and /v1/chat/completions endpoints."
      },
      "after": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for its Laguna models, and the models page says Laguna is free to use for a limited time. On 2026-10-04 that free offer covered both hosted models: Poolside states for Laguna XS 2.1 that free and paid endpoints are both available, with paid use priced at $0.10, $0.20 and $0.05 per 1M input, output and cache-read tokens. No quota for the free endpoints and no end date for the promotion are published.",
        "models_detail": "Poolside-hosted inference served two Laguna models on 2026-10-04. Laguna S 2.1, called with the model ID poolside/laguna-s-2.1, is a 118B total parameter mixture-of-experts model with 8B activated. Laguna XS 2.1 is a 33B total parameter mixture-of-experts model with 3B activated, served at 256K context on Poolside's API. Both are reached through an OpenAI-compatible chat API at https://inference.poolside.ai/v1, which exposes a model listing endpoint and a chat completions endpoint."
      }
    },
    {
      "id": "free-program-e12dfe20e799598fb5cd",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The 1M/250K group on 2026-10-02: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group on 2026-10-02: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the large-model list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "On 2026-10-04 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in its list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-bef2fba68857dcd93832",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "free_terms",
        "card",
        "data_treatment"
      ],
      "before": {
        "free_terms": "AI Horde is a standing free, community-hosted text and image inference service backed by volunteer compute. It permits anonymous API use with key 0000000000 and no registration, credit card, or paywall. Anonymous jobs receive the lowest queue priority; registered users can earn non-purchasable kudos by contributing compute, and capacity/model availability depends on active volunteer workers.",
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority.",
        "data_treatment": "Requests are processed by volunteer-operated workers around the world. The AI Horde Terms and Conditions and Privacy policy govern the service, including the OpenAI-compatible endpoint at oai.aihorde.net, and the public pages do not state how prompts or outputs are retained or used."
      },
      "after": {
        "free_terms": "AI Horde is a standing free, community-hosted text and image inference service backed by volunteer compute. It permits anonymous API use with the public key 0000000000 and requires no registration, credit card, or payment. Anonymous jobs receive the lowest queue priority, and the provider states that priority can only be earned by contributing compute and can never be bought or sold. Capacity and model availability depend on the volunteer workers connected at the time. An OpenAI-compatible endpoint at oai.aihorde.net also accepts the anonymous key at the lowest priority, and the provider describes that endpoint as a pilot that might be restricted or expanded in the future.",
        "card": "No credit card is required. The provider states that AI Horde can be used for free, even anonymously, with no credit card and no paywall, and that there is no way to buy priority.",
        "data_treatment": "Generation requests are fulfilled by workers running on computers that volunteers share, so prompts and outputs pass through machines operated by third parties. AI Horde publishes a privacy policy and terms of service, and the same terms and privacy policy apply to the OpenAI-compatible endpoint. The pages reviewed do not state whether prompts or outputs are retained or used to train models."
      }
    },
    {
      "id": "free-program-a529c3de2d1c8f72b7f7",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Eight catalog entries sit at $0.00/1M on 2026-10-02: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Upstage Solar Pro 4 is no longer in the free set and is priced at $0.09 in and $0.36 out per 1M on that date. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts, and the terms do not say which plans can enable Privacy Mode."
      },
      "after": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-10-04: inclusionAI Ling 3.0 Flash Fin, inclusionAI Ling 3.0 Flash Sante (free), inclusionAI Ling 3.1 Flash, Meituan LongCat 2.0, Meituan LongCat 2.5 Preview, Poolside Laguna S 2.1, Poolside Laguna XS 2.1, Space Bunny Alpha, and StepFun Step 3.7 Flash. Several of those names also appear as separate paid entries in the same catalog, so the $0.00/1M listing is the free one. Upstage Solar Pro 4 has left the free set and is now listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. The catalog is advertised at 300 or more models overall, and which entries are marked free can change.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it, and allow Nous to make Service Data and derived models available to third parties. The same terms state that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The terms do not say whether Privacy Mode is available on the $0 Free plan, and the plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-94bd3418d22fe342879a",
      "program_id": "catalog2-unorouter",
      "event_type": "program_record_corrected",
      "program_name": "UnoRouter free models",
      "provider_label": "UnoRouter",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://unorouter.com/en/docs/platform/models-and-pricing",
      "changed_fields": [
        "free_terms",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "UnoRouter offers models priced at $0 input and $0 output, marked with a :free suffix. A :free model routes only to free providers and never uses an account balance. A UnoRouter account and an API key are required to call them. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, or 503 when every provider for a model is busy. Some paid models are also offered free with a request size cap, and oversized prompts on that route return 413. There is no announced expiry, but availability is best-effort and the model set shifts over time.",
        "signup": "A UnoRouter account and an API key from the Tokens page are required to call the free models.",
        "clock": "Standing best-effort catalog. UnoRouter caps free models at 1 request per minute, per free model, per user, on top of upstream provider caps, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit on parallel requests. No expiry is published."
      },
      "after": {
        "free_terms": "UnoRouter publishes free models that carry a :free suffix, and a :free model routes only to free providers and never uses an account balance. Calling them requires a UnoRouter account and an API key. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, and 503 when every provider for a free model is busy. UnoRouter's published pages do not state which sign-in methods are offered or whether a payment method is required to open an account. There is no announced expiry, but availability is best-effort and the free model set shifts over time.",
        "signup": "A UnoRouter account and an API key are required. UnoRouter's published pages do not state which sign-in methods are offered or whether a payment method is needed at signup.",
        "clock": "Standing best-effort catalog with a cap of 1 request per minute, per free model, per user, plus the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit. No published expiry."
      }
    },
    {
      "id": "free-program-90e025432ab370e6dc0d",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "On 2026-10-02 the free allowance is 10 requests per minute and 60 weighted units per day for an account that holds no credit, and 20 requests per minute and 120 weighted units per day for an account that holds credit. The allowance is shared across the free models on an account rather than counted per model, with an additional IP guard, a site wide free cap of 15 requests per minute, and 2 concurrent free requests per account. The daily counter is measured in units that grow with prompt length, so long context requests use it up faster, and it resets at 00:00 UTC. Past the allowance, requests continue at the paid rate only when the account holds credit, and the X-Free-Fallback: false header turns that fallback off; an account with no balance gets 429 until the quota resets. The account rate limits in the documentation, 20 requests per minute below 5 US dollars of credit and 200 requests per minute at or above it, are account wide limits and not the free model allowance.",
        "models_detail": "Two free quota models on 2026-10-02: Qwen3.7 Flash (qwen/qwen3.7-flash) and DeepSeek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), plus auto:free routing, which always routes to a free model. The catalog changes over time.",
        "data_treatment": "Provider says it does not store message content by default and discards it after in memory processing, keeping billing metadata such as token counts, timestamps and model IDs. Every request passes automated content moderation, and the provider says it keeps the content of requests judged to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to their privacy policies."
      },
      "after": {
        "free_terms": "Accounts with no credit get 10 requests per minute and 60 weighted units per day; accounts that have topped up get 20 requests per minute and 120 weighted units per day. The free page states that this allowance is shared across the enabled free models, while the free-models documentation states that the daily quota is counted separately for each free model. Requests per minute use a rolling 60 second window and the daily counter resets at 00:00 UTC. A site-wide free cap of 15 requests per minute, a limit of 2 concurrent free requests per account, and an additional IP guard also apply. When weighted limits are enabled, longer inputs use more daily units, so long-context requests use the allowance up faster; otherwise each request counts once. Past the quota, requests continue at the paid rate only when paid fallback is enabled and the account holds credit, and sending the header X-Free-Fallback: false turns that fallback off; an account with no balance gets 429 until the quota resets. The documentation's example rate-limit header of 200 requests per minute is the paid account tier limit at $5 or more of credit, not the free quota.",
        "models_detail": "Two free-quota models were enabled on 2026-10-04: Qwen3.7 Flash (qwen/qwen3.7-flash), with a 1,000,000 token context and text, image and video input, and Deepseek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), with a 1,048,576 token context and text input. The auto:free route always resolves to a free model. Only enabled free models are listed, and the catalogue changes over time.",
        "data_treatment": "Provider says it does not store message content by default, processing it in memory and discarding it immediately, and that it keeps billing metadata such as token counts, timestamps and model IDs. Every request passes automated content moderation, and the provider says it retains the content of requests judged to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to those providers' privacy policies."
      }
    },
    {
      "id": "free-program-772ba741f7833d80bb0e",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-02) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per Kenari's homepage Solo plan listing on 2026-10-02 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog, and Kenari says the operator can change the quota, rate limit, and top-up threshold at any time).",
        "card": "No card is required for the Solo free lane, which Kenari lists as free with no top up. Its quickstart adds that no foreign credit card and no approval wait are needed, and optional top-up is by QRIS or crypto."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-04) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier card on Kenari's homepage on 2026-10-04 (the same counter read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required. Kenari's homepage lists the Solo tier as free with no top up, and its quickstart says no overseas credit card and no wait for approval are needed."
      }
    },
    {
      "id": "free-program-738eef32e4fc3d249e7d",
      "program_id": "routelm-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "RouteLM monthly free tokens",
      "provider_label": "RouteLM",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://routelm.dev/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "The 50,000 monthly free tokens apply only to the models the pricing table marks Eligible. On 2026-09-20 that list held eight models: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. Every named model outside that list (Claude, GPT, Grok, Kimi, MiniMax, GLM-5.2, DeepSeek V4 Pro) is marked Standard and is paid only. The auto smart router is listed separately with dynamic input and output costs and a free tier quota that varies by the model it routes to.",
        "signup": "required; account created with full name, email address, and password, followed by email verification and Turnstile CAPTCHA human verification"
      },
      "after": {
        "models_detail": "On 2026-10-04 the pricing table marks eight models Eligible for the 50,000 monthly free tokens: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. The other named models are marked Standard and are paid only: claude-opus-5, claude-sonnet-5, deepseek-v4-pro, glm-5.2, gpt-5.6-terra, gpt-5.6-sol, grok-4.5, kimi-k2.7-code, kimi-k3, and minimax-m2.7. The auto smart router is listed separately with dynamic pricing and a free tier quota that varies by the target model it selects, so an auto request routed to an eligible model can draw on the free allowance. The pricing page labels the catalog dynamic, so the roster can change.",
        "signup": "Required: create an account with your full name, email address, and a password. Signup is completed with email verification and Turnstile CAPTCHA human verification."
      }
    },
    {
      "id": "free-program-476702c1aa4bce21f7e6",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens, while agnes-2.5-pro is billed at its published Pro-tier prices. A new user can create an account and generate an API key without subscribing, then call the hosted chat completions API. Free and default text access is rate-limited to an effective 10 requests per minute, cut by 50 percent from the listed allowance of 30 in a change the provider dated September 23, 2026. The pricing page says offers can vary by model, account eligibility or promotion stage, and that the account bill is the source of truth for actual charges.",
        "models_detail": "On 2026-10-01 the pricing page lists agnes-2.5-flash and agnes-3.0-flash with cached input, input tokens and output tokens all at a current price of $0 per 1M, against list prices of $0.005, $0.05 and $0.15 per 1M. agnes-2.5-pro appears in the same table at its paid Pro-tier prices of $0.045 cached input, $0.45 input and $0.90 output per 1M. The text model table lists no other models."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to 10 requests per minute, after the provider cut free and enterprise text rate limits by 50 percent on 2026-09-23. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "On 2026-10-04 the text model pricing table lists agnes-2.5-flash and agnes-3.0-flash at $0 input, $0 output and $0 cached input, and the page states that cached input, input tokens and output tokens are currently free for both models. agnes-2.5-pro is billed at its published Pro-tier prices, and agnes-3.0-pro is marked coming soon at those same Pro-tier prices. agnes-2.0-flash is not listed on the pricing page."
      }
    },
    {
      "id": "free-program-0d1d2e45c5f49920f023",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-04T00:00:00+00:00",
      "date": "2026-10-04",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls, shared across all free models, with no credit card and no expiry. A one-time top-up of $1 or more switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day shared across all free models, reset every day. Some models carry a higher request weight and so allow fewer calls, with published caps as low as 10 per day and 1 per minute. Once a cap is reached the API returns 429 until the next minute or day begins.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page on 2026-10-01, nearly all text models plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The recurring daily quotas require a one-time top-up of $1 or more, which is a payment rather than a card kept on file, and free usage still bills $0 per token rather than drawing that balance down."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls across the free catalog with no credit card and no expiry. A one-time top-up of $1 or more then switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single shared pool of 1M tokens per day, all reset every day. Some models carry lower request caps because one call counts as several requests. On 2026-10-03 those caps were 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the Gemini image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors are listed at $0 input and $0 output on the provider's free catalog page, which was stamped Updated 2026-10-03 (read 2026-10-03). Most return text only, many also accept vision, audio, video, or PDF input, and two cover images: gpt-image-2-free returns text and images, and gemini-3.1-flash-image-preview-free returns images. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. A one-time top-up of $1 or more, which is a payment rather than a card kept on file, unlocks the daily free quotas across the free catalog."
      }
    },
    {
      "id": "free-event-d170846816db4de7ce9d",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "apodex/apodex-1.1-mini:free",
      "model": "Apodex: Apodex 1.1 Mini",
      "model_key": "apodex11minifree",
      "detected_at": "2026-10-02T11:30:12+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://openrouter.ai/apodex/apodex-1.1-mini:free"
    },
    {
      "id": "free-program-f1991410ce7324797e1f",
      "program_id": "morph-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Morph free tier (200 requests per month)",
      "provider_label": "Morph",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://www.morphllm.com/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Morph documents a recurring $0 free tier of 200 API requests per month. On 2026-09-25 the pricing page counts one request for each API call to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, or Rerank against that 200 per month cap, and states that per-token or plan-credit pricing applies once the cap is exhausted. The same page lists a Free subscription plan at $0 per month carrying 250K credits and low rate limits, with 1 credit worth about $0.00001 and credits applying to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, and Rerank. Morph's hosted chat models are listed with per-token rates, and the page does not state that the free request allowance or the free credits cover them.",
        "models_detail": "On 2026-09-25 the documented free allowance covers Fast Apply, WarpGrep, Compact, Reflex, Embeddings, and Rerank: 200 requests per month, plus 250K credits per month on the $0 Free plan. The hosted chat models listed on the pricing page that day, Qwen 3.8, GLM-5.3, GLM-5.3-Flash, and DeepSeek V4.1 Flash, carry per-token prices and are not named in the free-tier request or credit list."
      },
      "after": {
        "free_terms": "Morph documents a recurring $0 free tier of 200 API requests per month. On 2026-10-02 the pricing page states that each API call to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, or Rerank counts as one request against that 200 per month cap, and that once the cap is exhausted, per-token or plan-credit pricing applies. The same page lists a Free subscription plan at $0 per month carrying 250,000 credits with low rate limits, where one credit is about $0.00001 and credits apply to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, and Rerank. Credits do not roll over: they reset monthly on the subscription date. Morph's hosted chat models are listed with per-token prices rather than inside the free request allowance.",
        "models_detail": "On 2026-10-02 the free tier covers 200 API requests per month across Fast Apply (morph-v3-fast and morph-v3-large), WarpGrep, Compact, Reflex, Embeddings (morph-embedding-v4), and Rerank (morph-rerank-v4), plus 250,000 monthly credits on the $0 Free plan that apply to those same products. Morph's hosted chat models, including Kimi K3, GLM-5.3, GLM-5.3-Flash, DeepSeek V4.1 Flash, DeepSeek V4 Flash 0731, and Qwen 3.8, are priced per token."
      }
    },
    {
      "id": "free-program-de2b6d735c9740df7fe1",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Ten free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen prices ten models at Free for input and output tokens on its published pricing table. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-30: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, and Jev 1.13 Free were added by 2026-09-30. Union Alpha Free left the free table by 2026-09-30 and Muse Spark 1.2 Contributor Free left it by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Jev 1.13 Free is the free version of the Jev structured decision model and is called at the systemone endpoint rather than a chat endpoint. All other Zen models are pay-as-you-go.",
        "data_treatment": "Model specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve the model during their free periods. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA free trial endpoints: use is logged for security and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and are not used for model training. Prompts and other inputs to the Jev models are not used for training, and data is retained in accordance with TypeSafe AI's Privacy Policy.",
        "clock": "All ten free models are limited-time promotions, and no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table. Nine of them are Free for input, output, and cached reads, and Jev 1.13 Free is Free for input and output with no cached read price listed. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-10-02: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's Jev System One endpoint, which evaluates typed questions instead of generating text, and the two Nemotron entries are NVIDIA free trial endpoints. Union Alpha Free no longer appears on the free table on 2026-10-02, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the provider tells users not to submit personal or confidential data. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. The providers behind Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and do not use data for model training. Jev 1.13 Free runs on the Jev APIs, where prompts and other inputs are not used for training and data is retained under TypeSafe AI's privacy policy.",
        "clock": "All ten are limited-time promotions; no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-b158126bcef0d8ec3fb0",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. One top-up of $1 or more then switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free catalog and reset every day. Models with a higher request weight carry lower caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute, listed beside each model in the catalog table on 2026-10-01. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page, read on 2026-10-01 with the page stamped updated 2026-09-30. Almost all return text, many also accept vision, audio, video or PDF input, and two produce images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The daily free quotas require one top-up of $1 or more, a single payment rather than a card kept on file, and the reviewed pages do not state which payment methods that top-up accepts.",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls, shared across all free models, with no credit card and no expiry. A one-time top-up of $1 or more switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day shared across all free models, reset every day. Some models carry a higher request weight and so allow fewer calls, with published caps as low as 10 per day and 1 per minute. Once a cap is reached the API returns 429 until the next minute or day begins.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page on 2026-10-01, nearly all text models plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The recurring daily quotas require a one-time top-up of $1 or more, which is a payment rather than a card kept on file, and free usage still bills $0 per token rather than drawing that balance down.",
        "card_required": "yes"
      }
    },
    {
      "id": "free-program-aaefa39aa11b1792c5c9",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Current 1M/250K group on 2026-10-01: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-10-01: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The quota is shared across each group. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group list but marks it deprecated and shut down as of 7/14/25, so it is not current access. Fine-tuned models, fine-tuning training, evals, and tool use are excluded from the offer."
      },
      "after": {
        "models_detail": "The 1M/250K group on 2026-10-02: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M group on 2026-10-02: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the large-model list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-9d4622bc3a09d2f458f7",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-10-01: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Several of these names also appear as separate paid rows in the same catalog, so the $0.00/1M listing is the free one. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. The free set changes without notice.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The terms do not say whether Privacy Mode is limited to any particular plan, and the plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Eight catalog entries sit at $0.00/1M on 2026-10-02: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Upstage Solar Pro 4 is no longer in the free set and is priced at $0.09 in and $0.36 out per 1M on that date. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts, and the terms do not say which plans can enable Privacy Mode."
      }
    },
    {
      "id": "free-program-8e70aabadcda3cb6973e",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, counted per model; accounts that have topped up get twice the daily quota, and auto:free routing carries a separate parallel cap counted per IP address. The quota is counted in units that grow with prompt length, so long-context requests use it up faster. Past the quota, requests continue at the paid rate only when the account holds credit (a header can turn that fallback off); an account with no balance gets 429 until the quota resets. A separate account-wide rate limit also applies: 20 requests per minute while the account holds less than $5 of credit, and 200 requests per minute at $5 or more, with no daily cap at either level, so the example rate-limit header showing 200 requests per minute in the docs is the paid-tier figure.",
        "models_detail": "Two current free-quota models on 2026-09-17 (Qwen: Qwen3.7 Flash and DeepSeek: DeepSeek V4 Flash 0731free) plus auto:free routing; the catalog changes over time.",
        "data_treatment": "Provider says it does not store message content by default: content is processed in memory and then discarded, while token counts, timestamps and model IDs are kept for billing. Every request passes automated content moderation, and the provider keeps the content of requests it judges to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to their privacy policies."
      },
      "after": {
        "free_terms": "On 2026-10-02 the free allowance is 10 requests per minute and 60 weighted units per day for an account that holds no credit, and 20 requests per minute and 120 weighted units per day for an account that holds credit. The allowance is shared across the free models on an account rather than counted per model, with an additional IP guard, a site wide free cap of 15 requests per minute, and 2 concurrent free requests per account. The daily counter is measured in units that grow with prompt length, so long context requests use it up faster, and it resets at 00:00 UTC. Past the allowance, requests continue at the paid rate only when the account holds credit, and the X-Free-Fallback: false header turns that fallback off; an account with no balance gets 429 until the quota resets. The account rate limits in the documentation, 20 requests per minute below 5 US dollars of credit and 200 requests per minute at or above it, are account wide limits and not the free model allowance.",
        "models_detail": "Two free quota models on 2026-10-02: Qwen3.7 Flash (qwen/qwen3.7-flash) and DeepSeek V4 Flash 0731free (deepseek/deepseek-v4-flash-0731free), plus auto:free routing, which always routes to a free model. The catalog changes over time.",
        "data_treatment": "Provider says it does not store message content by default and discards it after in memory processing, keeping billing metadata such as token counts, timestamps and model IDs. Every request passes automated content moderation, and the provider says it keeps the content of requests judged to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to their privacy policies."
      }
    },
    {
      "id": "free-program-5223dd6ae6cf3a5edb43",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-01 the catalog listed 20 models, spanning embeddings, image generation, LLM guard, large language models, reasoning LLMs, speech to text, text to speech, and visual LLMs, drawn from the Llama, Qwen, Qwen3Guard, gpt-oss, bge, Whisper, NVIDIA Riva, and Stable Diffusion families. OVHcloud's product page describes the service as offering more than 20 models. Individual models can be retired under OVHcloud's lifecycle process. Seven catalog entries were priced at Free outright on 2026-10-01: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices. Qwen3.8-27B is paid at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-02 the catalog lists 20 models, covering Llama, Qwen, GPT-OSS, BGE embedding, Whisper speech-to-text, NVIDIA Riva text-to-speech, and Stable Diffusion image generation. Individual models can be retired under OVHcloud's lifecycle process, and a model announced for decommissioning stays callable through the API during its grace period while being hidden from the catalog, so the catalog count can understate what the API still serves. The authenticated catalog also prices several entries at Free outright, on 2026-10-02 the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices (Qwen3.8-27B itself is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens)."
      }
    },
    {
      "id": "free-program-3da603581ed3e95363a8",
      "program_id": "catalog2-unorouter",
      "event_type": "program_record_corrected",
      "program_name": "UnoRouter free models",
      "provider_label": "UnoRouter",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://unorouter.com/en/docs/platform/models-and-pricing",
      "changed_fields": [
        "free_terms",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "UnoRouter offers free model routes marked with a :free suffix and priced at $0 input and $0 output. A :free model routes only to free providers and never uses an account balance, while the same model name without the suffix is the paid version billed per token. Calling a free model requires a UnoRouter account and an API key. UnoRouter caps free use at 1 request per minute, per free model, per user, and further limits apply: the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429 with a Retry-After header, and 503 when every free provider for that model is busy. UnoRouter publishes no expiry for the free routes, but availability is best effort and a model leaves the catalog when every channel is out.",
        "signup": "A UnoRouter account and an API key are required, and UnoRouter's Quickstart covers getting a key. UnoRouter does not state which account sign-in methods it accepts, and it does not state whether a credit card is required.",
        "clock": "Standing best-effort catalog with no published expiry. Free use is capped at 1 request per minute, per free model, per user, layered with the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit."
      },
      "after": {
        "free_terms": "UnoRouter offers models priced at $0 input and $0 output, marked with a :free suffix. A :free model routes only to free providers and never uses an account balance. A UnoRouter account and an API key are required to call them. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, or 503 when every provider for a model is busy. Some paid models are also offered free with a request size cap, and oversized prompts on that route return 413. There is no announced expiry, but availability is best-effort and the model set shifts over time.",
        "signup": "A UnoRouter account and an API key from the Tokens page are required to call the free models.",
        "clock": "Standing best-effort catalog. UnoRouter caps free models at 1 request per minute, per free model, per user, on top of upstream provider caps, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit on parallel requests. No expiry is published."
      }
    },
    {
      "id": "free-program-38cff3327121ae7b21a6",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-10-01. Solar Pro 2 and Solar Pro 3 are no longer named in the program, and the API pricing page lists both with an end of service date of 2026-10-30 (KST).",
        "signup": "Application and approval required; eligible K-12 and university institutions, university hospitals, nonprofits, and NGOs. API access follows approval.",
        "clock": "The program window runs through 2027-03-31 23:00 KST. Approved organizations get access from the date of approval until 2027-03-31, for up to one year, so an approval later in the window yields less than a full year."
      },
      "after": {
        "models_detail": "Solar Pro 4 (LLM) and Document Parse, as named on the program page on 2026-10-02. The API pricing page lists the older Solar Pro 2 and Solar Pro 3 with an end of service date of 2026-10-30 KST.",
        "signup": "Application and approval required; eligible K-12 and university institutions, university hospitals, nonprofits, and NGOs. Applicants must create an Upstage Console account using the same email address before submitting the application form. Approved organizations receive setup instructions by email, and API access follows approval.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved access lasts up to one year, and the program page states that access runs from approval until 2027-03-31, so the one year is capped by that date."
      }
    },
    {
      "id": "free-program-36d01ae7714804b23f9c",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-10-01: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free-models table is shorter, listing nine of these and leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-10-02: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The Requesty documentation page carries a shorter table of nine of these models, leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan requires no credit card."
      }
    },
    {
      "id": "free-program-2366c5db117e98d33f6d",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-02T00:00:00+00:00",
      "date": "2026-10-02",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-01) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models covered by the Solo daily quota on 2026-10-01 per Kenari's homepage plan card (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "The Solo tier is advertised as free with no top-up, and Kenari's quickstart says no foreign credit card is needed and there is no approval wait, so no card is entered to reach the free models."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-02) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per Kenari's homepage Solo plan listing on 2026-10-02 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog, and Kenari says the operator can change the quota, rate limit, and top-up threshold at any time).",
        "card": "No card is required for the Solo free lane, which Kenari lists as free with no top up. Its quickstart adds that no foreign credit card and no approval wait are needed, and optional top-up is by QRIS or crypto."
      }
    },
    {
      "id": "free-program-cfb64ed735f8b5d99cc1",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card and no top up, with 50 requests per day across the current free-model roster (16 models on 2026-09-30) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models on the Solo tier per Kenari's homepage on 2026-09-30 (the homepage read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card and no top up are needed for the Solo free tier, which Kenari's homepage lists as free with no top up. Kenari's quickstart also states that no foreign credit card is needed, and optional top-ups go through QRIS from Rp 1.000."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-10-01) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models covered by the Solo daily quota on 2026-10-01 per Kenari's homepage plan card (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "The Solo tier is advertised as free with no top-up, and Kenari's quickstart says no foreign credit card is needed and there is no approval wait, so no card is entered to reach the free models."
      }
    },
    {
      "id": "free-program-b5c3034b7959398328b6",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens on 2026-09-29. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free and default text access is rate-limited to an effective 10 requests per minute, after a platform notice dated September 23, 2026 halved the text model rate limits for free and enterprise users. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "On 2026-09-29 the pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 input, $0 output and $0 cached input, and states that all three billing items are currently free for both models. agnes-2.5-pro appears on the same table at its paid Pro-tier prices. agnes-2.0-flash no longer appears on the pricing page."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens, while agnes-2.5-pro is billed at its published Pro-tier prices. A new user can create an account and generate an API key without subscribing, then call the hosted chat completions API. Free and default text access is rate-limited to an effective 10 requests per minute, cut by 50 percent from the listed allowance of 30 in a change the provider dated September 23, 2026. The pricing page says offers can vary by model, account eligibility or promotion stage, and that the account bill is the source of truth for actual charges.",
        "models_detail": "On 2026-10-01 the pricing page lists agnes-2.5-flash and agnes-3.0-flash with cached input, input tokens and output tokens all at a current price of $0 per 1M, against list prices of $0.005, $0.05 and $0.15 per 1M. agnes-2.5-pro appears in the same table at its paid Pro-tier prices of $0.045 cached input, $0.45 input and $0.90 output per 1M. The text model table lists no other models."
      }
    },
    {
      "id": "free-program-99223663f2e7814da153",
      "program_id": "catalog2-orcarouter",
      "event_type": "program_terms_changed",
      "program_name": "OrcaRouter free models",
      "provider_label": "OrcaRouter",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://docs.orcarouter.ai/routing/free-models",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "OrcaRouter provides orcarouter/free and explicit -free model IDs that never touch the account wallet and can be used indefinitely, subject to request-rate, daily, and per-request prompt-size limits. The per-minute and per-day request ceilings for every tier are published at GET /api/free-package/public and are tuned live, and the day window rolls at 00:00 UTC. The per-request prompt cap that applies on the lowest tier is the one limit not published as a number. Free limits are tiered by lifetime spend: an account that has never topped up gets a deliberately small daily allowance, and crossing the lifetime-spend threshold moves the account to a much larger daily budget. The general orcarouter/auto router can resolve to ordinary paid models billed at upstream prices, so it is not part of this free program.",
        "models_detail": "orcarouter/free plus the explicit -free model IDs, which are marked by the -free suffix and a zero-price block in /v1/models and the public pricing catalog. The free lineup rotates and the documentation deliberately does not list the IDs, so the current set is read from GET /api/free-package/public. On 2026-09-30 the orcarouter/free router page listed one model in the router, deepseek/deepseek-v4-flash-free, on the light tier. The orcarouter/auto alias can select paid models and is not part of this free program.",
        "signup": "OrcaRouter account and API key required, and free routes never touch the wallet. Free access also requires the workspace owner to have linked a GitHub account that has been registered for some time, unless the workspace has paid enough to be exempt. A workspace without one is refused on free models with error code free_rate_limited and reason err_free_access_denied, and the documented remedy is to link an established GitHub account in the owner's profile settings or to add credits.",
        "clock": "Standing free route with no stated end date. Request-rate and daily ceilings are tuned live and published per tier at GET /api/free-package/public, with the day window rolling at 00:00 UTC. The per-request prompt cap on the lowest tier is not published as a number."
      },
      "after": {
        "free_terms": "OrcaRouter serves selected catalog models at $0 under model ids ending in -free, and the built-in orcarouter/free router covers that free tier behind a single id. Calls to these ids never touch the account wallet, and the allowance is a request rate rather than a balance, so a workspace can keep calling the free tier indefinitely while it stays inside the limits. Free traffic is capped by requests per minute and requests per day. Those ceilings are public, returned for every tier by the GET /api/free-package/public endpoint, and the provider tunes them live. They are tiered by lifetime spend: a workspace that has never topped up gets a deliberately small daily allowance and also a per-request prompt-size cap, which is the one limit the provider does not publish as a number, and crossing the lifetime-spend threshold raises both. Paying is optional and only buys higher limits. Free access additionally requires the workspace owner to have linked a GitHub account that has been registered for some time, unless the workspace has paid enough to be exempt. Model ids without the -free suffix bill at the upstream per-token price, and a free id never falls back to a paid model.",
        "models_detail": "The free lineup is the catalog ids ending in -free plus the orcarouter/free router, which scores each request by difficulty and picks among the workspace's free models. The provider states that the lineup rotates as models join and leave with capacity, so its documentation deliberately lists no ids: the current ids, and the paid model each one shadows, come from the public free-package endpoint. On 2026-10-01 the orcarouter/free model page listed one model in the router, deepseek/deepseek-v4-flash-free, on the light tier.",
        "signup": "An OrcaRouter account and API key are required, and free routes never touch the wallet. Free access is also gated on the workspace owner having linked a GitHub account that has been registered for some time, unless the workspace has paid enough to be exempt. A workspace that meets neither condition has its free requests refused.",
        "clock": "Standing free route with no stated end date. The request-per-minute and request-per-day ceilings are published per tier at the public free-package endpoint and are tuned live, and the lower tier also carries a per-request prompt-size cap that the provider does not publish as a number. The provider describes the free tier as best-effort capacity rather than a guarantee."
      }
    },
    {
      "id": "free-program-90de05e4cf4c82a4b070",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-30: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free models table is shorter, listing nine of these and omitting Nemotron 3.5 Lightning, Ling 3.0 Tiny, and Muse Glimmer 30B. Requesty says these models are free for now, that pricing may change in the future, and that any change will be announced in the changelog ahead of time.",
        "data_treatment": "Requesty says it never uses your data for training, and that each model detail page shows the data retention and training policy of the provider serving that model, so treatment varies by route. Check the model detail page and the underlying provider's policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-10-01: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free-models table is shorter, listing nine of these and leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer 30B. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policies for the underlying model providers vary, and each model's detail page shows the policy that applies to that provider, so check them before sensitive use."
      }
    },
    {
      "id": "free-program-86f5c7abad88269b4858",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Models whose pricing tables mark standard Free tier input and output as free of charge, as listed on 2026-09-30, are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries marked free of charge on 2026-09-04: Gemini 3.5 Transcribe Live for streaming and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first-generation Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview no longer have pricing tables on the page and are no longer part of the free roster. This is model-specific: other listed models, including the Pro Preview, image, video, and music models, show Free tier \"Not available\", and several batch and flex modes are also unavailable to the Free tier."
      },
      "after": {
        "models_detail": "On 2026-10-01 the models whose pricing tables mark Standard Free tier input and output as free of charge are Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries, Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first-generation Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview are no longer listed on the pricing page. This is model-specific: other listed models, including the Pro Preview, image, video, and music models, show Free tier 'Not available', and some batch, flex, and priority modes do the same."
      }
    },
    {
      "id": "free-program-7ce9a5a0bd37ac988169",
      "program_id": "relace-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "Relace model API free tier",
      "provider_label": "Relace",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://relace.ai/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Relace offers a hosted API free tier for testing its code-focused models (Search, Compact, Apply, and Rank). Signup is required. The public pricing page does not disclose the free quota, renewal cadence, expiration, or whether a payment method is required.",
        "models_detail": "Relace model API free tier",
        "signup": "Relace account and API key required; current provider material does not state a card or funded balance requirement for the free model tier."
      },
      "after": {
        "free_terms": "Relace offers a hosted API free tier for testing its code-focused models. Signup is required. The API reference states that rate limits are a per-key request budget over a rolling 60 second window, that free accounts start at 15 requests per minute, and that the budget scales with account tier. No free token or request quota, renewal cadence, or expiration date is published.",
        "models_detail": "On 2026-10-01 the pricing page lists four Relace models with per-token prices: relace-search, relace-compact, relace-apply-3, and relace-rank. The API reference also documents Open Models endpoints for chat completions, messages, and listing models. Provider material does not state which of these models the free tier covers.",
        "signup": "A Relace account and an API key carrying the rlc- prefix are required. Provider material does not state that the free tier itself requires a card or a funded balance, though the API reference lists a 402 response for an account that is out of credits or has no payment method attached."
      }
    },
    {
      "id": "free-program-57adc30e7f07f1eeba9f",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of at least $1 permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, reset every day. Models that carry a higher request weight have lower caps, and on 2026-09-30 those ran at 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the Gemini image preview model. Past a cap the API returns a 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page (read on 2026-09-30, page stamped Updated 2026-09-29), nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Reaching the recurring daily quotas requires a one-time top-up of at least $1, and free usage bills $0 rather than drawing that balance down. The provider pages do not state which payment methods the top-up accepts."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. One top-up of $1 or more then switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free catalog and reset every day. Models with a higher request weight carry lower caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute, listed beside each model in the catalog table on 2026-10-01. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page, read on 2026-10-01 with the page stamped updated 2026-09-30. Almost all return text, many also accept vision, audio, video or PDF input, and two produce images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The daily free quotas require one top-up of $1 or more, a single payment rather than a card kept on file, and the reviewed pages do not state which payment methods that top-up accepts."
      }
    },
    {
      "id": "free-program-497d9e8599728142ef3f",
      "program_id": "llm7-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "LLM7 free-token tier",
      "provider_label": "LLM7.io",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://docs.llm7.io/limits",
      "changed_fields": [
        "free_terms",
        "signup",
        "signup_required"
      ],
      "before": {
        "free_terms": "Anonymous: 10 RPM, 60/hour, 500K tokens/24h. Free token: 40 RPM, 100/hour, 1M tokens/24h.",
        "signup": "No account for anonymous access; free dashboard token for higher limits.",
        "signup_required": "no"
      },
      "after": {
        "free_terms": "Free token access allows 1 request per second, 60 requests per minute, and 250 requests per hour on text generation, with 100,000 tokens per 24 hours counted as input tokens plus output tokens. The provider states that free-token quotas are provided at no charge and may be reduced without notice.",
        "signup": "A free API token from the LLM7 dashboard at dash.llm7.io is required. As of 2026-10-01 the provider documents only two access types, Free token and Pro, so there is no documented anonymous access.",
        "signup_required": "yes"
      }
    },
    {
      "id": "free-program-35ddeede4f5eb7b93c87",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "On 2026-09-30 the program page lists free access to Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse. Solar Pro 2 is no longer named in the program. Upstage's API pricing page states that Solar Pro 3 reaches end of service on 2026-10-30 (KST).",
        "clock": "Applications are accepted through 2027-03-31 11:00 PM (KST). Approved access runs from the date of approval until 2027-03-31 and lasts up to one year."
      },
      "after": {
        "models_detail": "Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-10-01. Solar Pro 2 and Solar Pro 3 are no longer named in the program, and the API pricing page lists both with an end of service date of 2026-10-30 (KST).",
        "clock": "The program window runs through 2027-03-31 23:00 KST. Approved organizations get access from the date of approval until 2027-03-31, for up to one year, so an approval later in the window yields less than a full year."
      }
    },
    {
      "id": "free-program-352f8f2e2dcc52ddca3f",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide, read on 2026-09-30, states that API key authentication is free of charge with daily limits, while AWS IAM authentication instead applies usage charges to an AWS account, so the zero-cost route is the API key issued from the Nova developer surface rather than AWS IAM access. No numeric limits are published publicly."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is free of charge with daily limits, while the separate AWS IAM authentication route instead applies usage charges to an AWS account. No numeric limits are published publicly."
      }
    },
    {
      "id": "free-program-3520437079edc0c55f38",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-09-30: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Several of those names also appear separately in the catalog at regular paid prices. Tencent Hy3 is not in the free set and lists at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-10-01: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash. Several of these names also appear as separate paid rows in the same catalog, so the $0.00/1M listing is the free one. Upstage Solar Pro 4 has left the free set and is listed at $0.09 in and $0.36 out per 1M. Tencent Hy3 is also outside the free set, at $0.13 in and $0.53 out per 1M. The free set changes without notice.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the Privacy Policy. The terms do not say whether Privacy Mode is limited to any particular plan, and the plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-18f0aafd011672e1100b",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "As of 2026-09-30 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. As of the same date the 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "Current 1M/250K group on 2026-10-01: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-10-01: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The quota is shared across each group. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group list but marks it deprecated and shut down as of 7/14/25, so it is not current access. Fine-tuned models, fine-tuning training, evals, and tool use are excluded from the offer."
      }
    },
    {
      "id": "free-program-18e14322fc2d1cef317d",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-10-01T00:00:00+00:00",
      "date": "2026-10-01",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-30 the catalog listed 20 models, including Llama, Qwen, GPT-OSS, Whisper speech to text, bge and Qwen3 embeddings, NVIDIA Riva text to speech, and Stable Diffusion XL image generation. Individual models can be retired under OVHcloud's lifecycle process. Seven catalog entries were priced at Free outright on 2026-09-30: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text to speech voices for German, Spanish, Italian, and US English. Other entries are billed per million tokens, for example Qwen3.8-27B at 0.4 EUR input and 2.7 EUR output per 1M tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-10-01 the catalog listed 20 models, spanning embeddings, image generation, LLM guard, large language models, reasoning LLMs, speech to text, text to speech, and visual LLMs, drawn from the Llama, Qwen, Qwen3Guard, gpt-oss, bge, Whisper, NVIDIA Riva, and Stable Diffusion families. OVHcloud's product page describes the service as offering more than 20 models. Individual models can be retired under OVHcloud's lifecycle process. Seven catalog entries were priced at Free outright on 2026-10-01: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices. Qwen3.8-27B is paid at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens."
      }
    },
    {
      "id": "free-program-f1407ca39f980ff3f144",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-29 the 1M/250K token group listed: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M token group listed: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also keeps gpt-4.5-preview-2025-02-27 in the 1M group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "As of 2026-09-30 the 1M/250K group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. As of the same date the 10M/2.5M group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-eb9a3a90b1903b8d81d1",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Ten free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table, covering input and output tokens and, where a cached read price is listed, cached reads as well. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-29: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model called through the System One endpoint rather than a text generator. Union Alpha Free left the free table by 2026-09-29. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free are served by providers that follow a zero-retention policy and do not use your data for model training; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA free endpoints for trial use only, and use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and do not use your data for model training. Jev prompts and other inputs are not used for training, and data is retained in accordance with TypeSafe AI's privacy policy."
      },
      "after": {
        "free_terms": "OpenCode Zen prices ten models at Free for input and output tokens on its published pricing table. The provider describes every free model as available for a limited time and publishes no request quota and no end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-30: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, and Jev 1.13 Free were added by 2026-09-30. Union Alpha Free left the free table by 2026-09-30 and Muse Spark 1.2 Contributor Free left it by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Jev 1.13 Free is the free version of the Jev structured decision model and is called at the systemone endpoint rather than a chat endpoint. All other Zen models are pay-as-you-go.",
        "data_treatment": "Model specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve the model during their free periods. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA free trial endpoints: use is logged for security and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and are not used for model training. Prompts and other inputs to the Jev models are not used for training, and data is retained in accordance with TypeSafe AI's Privacy Policy."
      }
    },
    {
      "id": "free-program-d634bec7fe3675027fa2",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset every day. Models that carry a higher request weight have lower request caps, as low as 10 per day and 1 per minute, and the free catalog table lists the cap for each model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models were listed at $0 input and $0 output on the provider's free catalog page on 2026-09-29, from 16 model authors. Nearly all are text models, plus two that handle images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Unlocking the daily quotas requires a one-time top-up of any amount from $1, and the provider does not state which payment methods it accepts for that top-up. Free models bill $0 within the quotas, so free usage does not draw the topped-up balance down."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of at least $1 permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, reset every day. Models that carry a higher request weight have lower caps, and on 2026-09-30 those ran at 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the Gemini image preview model. Past a cap the API returns a 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page (read on 2026-09-30, page stamped Updated 2026-09-29), nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Reaching the recurring daily quotas requires a one-time top-up of at least $1, and free usage bills $0 rather than drawing that balance down. The provider pages do not state which payment methods the top-up accepts."
      }
    },
    {
      "id": "free-program-d30671c2929e72d00ebb",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page (real-time price sync, available models only), usable within platform rate limits. Identity verification is required first: the provider states that after verifying your identity you can use all free models, and that free model calls are free and appear as a cost of 0 in the account bill. Rate limits for free models are fixed values, and per-model rate limits are shown on each model detail page after login. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-29, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B (ChinaTelecom, 29B total parameters with 4B active and 256K native context, listed at \uffe50 input and \uffe50 output and released on 2026-09-17), THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), four BAAI bge embedding and reranker models, six speech models including XingChenAGI/XingChenASR-V3.2-Ultra, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall, and one image model, Kwai-Kolors/Kolors. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Identity verification is also required: the provider states that after verifying your identity you can use all free models."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page, which presents itself as showing real-time prices and available models only, usable within platform rate limits. Rate limits for free models are fixed values, and free model calls appear in the account bill at a cost of 0. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-30, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and XingChenAGI/Xing4.0-29B from China Telecom. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is listed.",
        "signup": "required; account at cloud.siliconflow.cn, with the platform supporting login by SMS and by email, and the API key created in the console. The platform documentation states that after verifying your identity you can use all free models, so identity verification is part of reaching the free roster."
      }
    },
    {
      "id": "free-program-bc04376a470a958ff6c1",
      "program_id": "catalog2-merge-gateway",
      "event_type": "program_terms_changed",
      "program_name": "Merge Gateway free Nemotron route",
      "provider_label": "Merge Gateway",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://docs.merge.dev/merge-gateway/models/details/nvidia-nemotron-3-5-lightning-30b-a3b",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Merge's provider-owned catalog lists nvidia/nemotron-3.5-lightning-30b-a3b as available hosted text inference with one NVIDIA host and Free / Free input and output pricing, checked on 2026-09-23. Merge's Gateway pricing page puts access to all free LLMs, automatic routing with built-in fallback, and a start without a credit card in the Free plan, and lists all free models as included in every plan. The Free plan column shows no credit amount, while Merge's Get started documentation says the free plan comes with credit to spend on its model list, so treat any plan credit as an extra rather than as the thing paying for this route. Pro is a paid tier that charges LLM cost plus a 5% fee and issues $10 in complimentary credits each month that expire at the end of the month. Merge also warns that the usage.cost field in a response is provider cost rather than the invoiced amount, but this exact Nemotron route is cataloged as free rather than merely lacking pricing metadata."
      },
      "after": {
        "free_terms": "Merge's provider-owned catalog lists nvidia/nemotron-3.5-lightning-30b-a3b as available hosted text inference with one NVIDIA host and Free / Free input and output pricing, as recorded on 2026-09-30. Merge's Free plan (pricing page, 2026-09-04) includes access to all free LLMs, automatic routing with built-in fallback, and starts without a credit card; the earlier $10 first-month credit was removed from the Free tier, while Pro adds $10 in expiring monthly credits and ordinarily charges model cost plus 5%. Merge's Gateway get started page adds that the Free plan covers its list of free models up to a daily request cap, and that adding a card moves the account to Pro and the full catalog. Merge does not publish the size of that daily cap on any of these pages. Merge also warns that its response usage.cost is provider cost rather than necessarily the final invoice, but the exact Nemotron route is explicitly cataloged as free rather than merely lacking pricing metadata."
      }
    },
    {
      "id": "free-program-909ba600cfed29174b16",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-28 the catalog listed 20 models drawn from the Qwen, Qwen3Guard, gpt-oss, Llama, Whisper, bge, Stable Diffusion, and NVIDIA Riva series, and OVHcloud describes the range as more than 20 models. Seven of those entries are priced at Free outright: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the NVIDIA Riva voices nvr-tts-de-de, nvr-tts-en-us, nvr-tts-es-es, and nvr-tts-it-it. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens. Individual models can be retired under OVHcloud's lifecycle process. On 2026-09-28 Mistral-7B-Instruct-v0.3 and Qwen3-Coder-30B-A3B-Instruct-FP8 were listed for decommissioning on 2026-10-01, and Mistral-Nemo-Instruct-2407 and Mistral-Small-3.2-24B-Instruct-2506 on 2026-11-04."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-30 the catalog listed 20 models, including Llama, Qwen, GPT-OSS, Whisper speech to text, bge and Qwen3 embeddings, NVIDIA Riva text to speech, and Stable Diffusion XL image generation. Individual models can be retired under OVHcloud's lifecycle process. Seven catalog entries were priced at Free outright on 2026-09-30: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text to speech voices for German, Spanish, Italian, and US English. Other entries are billed per million tokens, for example Qwen3.8-27B at 0.4 EUR input and 2.7 EUR output per 1M tokens."
      }
    },
    {
      "id": "free-program-562781162544233d0311",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Amazon Nova API key access is free of charge with daily limits. The Amazon Nova Act user guide states that API key authentication, using a key issued at nova.amazon.com/act and tied to an Amazon.com account, is free of charge with daily limits; the size of those daily limits is not published. A separate AWS IAM authentication route exists for production work and bills usage charges to an AWS account, so paying is an option for scale, not a condition of the free route. The Nova.amazon.com Terms of Use (last updated on December 2, 2025) contain no billing or fee language: API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the offering does not support deploying production scale Developer Offerings (S-1.1)."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide, read on 2026-09-30, states that API key authentication is free of charge with daily limits, while AWS IAM authentication instead applies usage charges to an AWS account, so the zero-cost route is the API key issued from the Nova developer surface rather than AWS IAM access. No numeric limits are published publicly."
      }
    },
    {
      "id": "free-program-489fbdbec9413cf07674",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "No credit card is required at any point. AI Horde states that everyone can use the service for free, even anonymously, with no credit card and no paywall."
      },
      "after": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority."
      }
    },
    {
      "id": "free-program-42460c417b4f526fb1b0",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-29. Solar Pro 2 is not named in the program.",
        "clock": "The program runs until 2027-03-31 at 11:00 PM KST. Approved organizations can use the free access from the date of approval until 2027-03-31, for up to one year, so an approval granted later in the window gives less than a full year."
      },
      "after": {
        "models_detail": "On 2026-09-30 the program page lists free access to Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse. Solar Pro 2 is no longer named in the program. Upstage's API pricing page states that Solar Pro 3 reaches end of service on 2026-10-30 (KST).",
        "clock": "Applications are accepted through 2027-03-31 11:00 PM (KST). Approved access runs from the date of approval until 2027-03-31 and lasts up to one year."
      }
    },
    {
      "id": "free-program-34bd0440d9bcbc8b8b80",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-29) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per Kenari's homepage Solo plan card on 2026-09-29 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required for the Solo free lane: the tier is listed as free with no top up, and Kenari's quickstart says no foreign credit card and no approval wait are needed."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card and no top up, with 50 requests per day across the current free-model roster (16 models on 2026-09-30) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models on the Solo tier per Kenari's homepage on 2026-09-30 (the homepage read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card and no top up are needed for the Solo free tier, which Kenari's homepage lists as free with no top up. Kenari's quickstart also states that no foreign credit card is needed, and optional top-ups go through QRIS from Rp 1.000."
      }
    },
    {
      "id": "free-program-26c2d9e7885a296b2245",
      "program_id": "catalog2-orcarouter",
      "event_type": "program_record_corrected",
      "program_name": "OrcaRouter free models",
      "provider_label": "OrcaRouter",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://docs.orcarouter.ai/routing/free-models",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "OrcaRouter provides orcarouter/free and explicit -free model IDs that never touch the account wallet and can be called indefinitely within rate limits. Requests are capped per minute and per day, and the day window resets at 00:00 UTC. The rpm and rpd numbers for each tier are published at GET /api/free-package/public and are tuned live. Limits scale with the account's lifetime spend: an account that has never topped up gets a small daily allowance. Lower tiers also have a per-request prompt-size cap, and that number is not published. The general orcarouter/auto router can resolve to ordinary paid models billed at upstream prices, so it is not part of this free program.",
        "models_detail": "orcarouter/free plus the explicit -free model IDs, which rotate as capacity allows. On 2026-09-17, orcarouter/free listed one model: deepseek/deepseek-v4-flash-free. The orcarouter/auto alias can select paid models and is not part of this free program.",
        "signup": "An OrcaRouter account and API key are required. Free requests can be refused unless the workspace owner links a GitHub account that has been registered for some time, or the workspace has paid enough to be exempt. Free routes never touch the wallet.",
        "clock": "Standing free route. Per-minute and per-day request limits are published per tier and tuned live, and the per-request prompt-size cap is unpublished."
      },
      "after": {
        "free_terms": "OrcaRouter provides orcarouter/free and explicit -free model IDs that never touch the account wallet and can be used indefinitely, subject to request-rate, daily, and per-request prompt-size limits. The per-minute and per-day request ceilings for every tier are published at GET /api/free-package/public and are tuned live, and the day window rolls at 00:00 UTC. The per-request prompt cap that applies on the lowest tier is the one limit not published as a number. Free limits are tiered by lifetime spend: an account that has never topped up gets a deliberately small daily allowance, and crossing the lifetime-spend threshold moves the account to a much larger daily budget. The general orcarouter/auto router can resolve to ordinary paid models billed at upstream prices, so it is not part of this free program.",
        "models_detail": "orcarouter/free plus the explicit -free model IDs, which are marked by the -free suffix and a zero-price block in /v1/models and the public pricing catalog. The free lineup rotates and the documentation deliberately does not list the IDs, so the current set is read from GET /api/free-package/public. On 2026-09-30 the orcarouter/free router page listed one model in the router, deepseek/deepseek-v4-flash-free, on the light tier. The orcarouter/auto alias can select paid models and is not part of this free program.",
        "signup": "OrcaRouter account and API key required, and free routes never touch the wallet. Free access also requires the workspace owner to have linked a GitHub account that has been registered for some time, unless the workspace has paid enough to be exempt. A workspace without one is refused on free models with error code free_rate_limited and reason err_free_access_denied, and the documented remedy is to link an established GitHub account in the owner's profile settings or to add credits.",
        "clock": "Standing free route with no stated end date. Request-rate and daily ceilings are tuned live and published per tier at GET /api/free-package/public, with the day window rolling at 00:00 UTC. The per-request prompt cap on the lowest tier is not published as a number."
      }
    },
    {
      "id": "free-program-13eb9de012665c91caf4",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Nine catalog entries were listed at $0.00 per 1M on 2026-09-29: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Several of these models also appear separately in the catalog at paid rates. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the privacy policy. The plan page itself says nothing more specific about free-inference prompts."
      },
      "after": {
        "models_detail": "Nine catalog entries were listed at $0.00/1M on 2026-09-30: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Several of those names also appear separately in the catalog at regular paid prices. Tencent Hy3 is not in the free set and lists at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement, or support purposes, except as described in the Privacy Policy. The plan page itself says nothing more specific about free-inference prompts."
      }
    },
    {
      "id": "free-program-0e74615c513d757e90b0",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "On 2026-09-29 the Gemini Developer API pricing tables mark standard Free tier input and output as free of charge for Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. The pricing page no longer carries entries for Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini 2.5 Flash-Lite Preview, the earlier Gemini Embedding model, or Gemini Robotics ER 1.6 Preview. Free access is model-specific: Gemini 3.1 Pro Preview, the Nano Banana image models, Gemini Omni Flash, Gemini 2.5 Computer Use Preview, Veo 3.1, and the Lyria music models show Free tier \"Not available\", and several batch and flex modes do the same.",
        "signup": "required; a Google account, active project, and Gemini API key are required"
      },
      "after": {
        "models_detail": "Models whose pricing tables mark standard Free tier input and output as free of charge, as listed on 2026-09-30, are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries marked free of charge on 2026-09-04: Gemini 3.5 Transcribe Live for streaming and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first-generation Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview no longer have pricing tables on the page and are no longer part of the free roster. This is model-specific: other listed models, including the Pro Preview, image, video, and music models, show Free tier \"Not available\", and several batch and flex modes are also unavailable to the Free tier.",
        "signup": "required; a Google account, an active project, and a Gemini API key are required, and the Additional Terms require users to be 18 years of age or older"
      }
    },
    {
      "id": "free-program-0077dbf8754eb9a53ea7",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-30T00:00:00+00:00",
      "date": "2026-09-30",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-29: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free-models table is shorter and lists nine of them, leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card is required. The pricing page lists \"No credit card required\" among the Free plan's terms.",
        "data_treatment": "Requesty says it never uses your data for training, and that each model detail page shows the data retention and training policy of the provider serving that model. Those upstream policies differ, so check the model detail page and the underlying provider's policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-30: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free models table is shorter, listing nine of these and omitting Nemotron 3.5 Lightning, Ling 3.0 Tiny, and Muse Glimmer 30B. Requesty says these models are free for now, that pricing may change in the future, and that any change will be announced in the changelog ahead of time.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses your data for training, and that each model detail page shows the data retention and training policy of the provider serving that model, so treatment varies by route. Check the model detail page and the underlying provider's policy before sensitive use."
      }
    },
    {
      "id": "free-event-62d62477674472e70423",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "inclusionai/ling-3.0-flash-fin:free",
      "model": "inclusionAI: Ling 3.0 Flash Fin",
      "model_key": "ling30flashfinfree",
      "detected_at": "2026-09-29T11:30:25+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-fin:free"
    },
    {
      "id": "free-program-ea0a05d7b949ece10ae0",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "On the Gemini API pricing page as of 2026-09-25, the models whose Standard tables mark Free tier input and output as free of charge are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live for streaming transcription, Gemini 3.5 Transcribe for diarizing transcription, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview are no longer listed on the pricing page. This is model specific: other listed models, including the Omni Flash, Pro, image, video, and music entries, show Free tier as Not available, and the batch, flex, and priority modes differ by model.",
        "clock": "not_stated; no fixed Free-tier end date is stated, and daily quotas reset according to the applicable limit"
      },
      "after": {
        "models_detail": "On 2026-09-29 the Gemini Developer API pricing tables mark standard Free tier input and output as free of charge for Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. The pricing page no longer carries entries for Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini 2.5 Flash-Lite Preview, the earlier Gemini Embedding model, or Gemini Robotics ER 1.6 Preview. Free access is model-specific: Gemini 3.1 Pro Preview, the Nano Banana image models, Gemini Omni Flash, Gemini 2.5 Computer Use Preview, Veo 3.1, and the Lyria music models show Free tier \"Not available\", and several batch and flex modes do the same.",
        "clock": "No end date is stated for the Free tier. Requests per day quotas reset at midnight Pacific time, and other limits are enforced per minute or per day depending on the model."
      }
    },
    {
      "id": "free-program-e377928085945a570851",
      "program_id": "poolside-laguna-limited-free",
      "event_type": "program_record_corrected",
      "program_name": "Laguna hosted API, free for a limited time",
      "provider_label": "Poolside",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://poolside.ai/models",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for Laguna models and says free and paid API endpoints are both available. The models page says hosted use of Laguna S 2.1 is free for a limited time; no free-endpoint quota or end date is disclosed.",
        "models_detail": "Laguna hosted API free for a limited time"
      },
      "after": {
        "free_terms": "Poolside offers provider-hosted, OpenAI-compatible chat inference for its Laguna models, and the models page says Laguna is free to use for a limited time. On 2026-09-29 that free section covers both Laguna S 2.1 and Laguna XS 2.1, and the Laguna XS 2.1 announcement says free and paid endpoints are both available, with paid usage priced at $0.10 per 1M input tokens, $0.20 per 1M output tokens, and $0.05 per 1M cache-read tokens. No quota or rate limit for the free endpoints is published, and no end date is given.",
        "models_detail": "Hosted Laguna models are free to use for a limited time. On 2026-09-29 the free offer covers Laguna S 2.1, called as poolside/laguna-s-2.1, and Laguna XS 2.1, both served over an OpenAI-compatible chat API at https://inference.poolside.ai/v1 with the /v1/models and /v1/chat/completions endpoints."
      }
    },
    {
      "id": "free-program-b1f01418e8e37f77beb5",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). The Amazon Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is free of charge with daily limits, while the alternative AWS IAM authentication path bills usage to an AWS account. No numeric limits are published publicly.",
        "card": "Not required. The Nova Act user guide states that API key authentication is free of charge with daily limits, and no payment terms appear anywhere in the Nova Terms of Use."
      },
      "after": {
        "free_terms": "Amazon Nova API key access is free of charge with daily limits. The Amazon Nova Act user guide states that API key authentication, using a key issued at nova.amazon.com/act and tied to an Amazon.com account, is free of charge with daily limits; the size of those daily limits is not published. A separate AWS IAM authentication route exists for production work and bills usage charges to an AWS account, so paying is an option for scale, not a condition of the free route. The Nova.amazon.com Terms of Use (last updated on December 2, 2025) contain no billing or fee language: API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the offering does not support deploying production scale Developer Offerings (S-1.1).",
        "card": "Not required. Amazon states that API key authentication is free of charge with daily limits, and the Nova.amazon.com Terms of Use contain no billing or fee language. Charges arise only on the optional AWS IAM authentication route, which bills an AWS account."
      }
    },
    {
      "id": "free-program-ada7fedf2410adbf5db4",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-28 the 1M/250K group listed gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group listed gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The quota is shared across each group. The page also retains gpt-4.5-preview-2025-02-27 in the 1M list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "On 2026-09-29 the 1M/250K token group listed: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M token group listed: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also keeps gpt-4.5-preview-2025-02-27 in the 1M group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-a3006cb83cbd4b76bab4",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-28. Solar Pro 2 is no longer named in the program.",
        "clock": "The program runs through 2027-03-31 23:00 KST. Approved organizations have access from the date their application is approved until 2027-03-31, for up to one year."
      },
      "after": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-29. Solar Pro 2 is not named in the program.",
        "clock": "The program runs until 2027-03-31 at 11:00 PM KST. Approved organizations can use the free access from the date of approval until 2027-03-31, for up to one year, so an approval granted later in the window gives less than a full year."
      }
    },
    {
      "id": "free-program-9f7b4a97e0f949d7486d",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Ten free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices ten models at Free for input and output tokens, and nine of those also show Free for cached reads. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-28: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's structured decision model and is called at the System One endpoint. Union Alpha Free no longer appears on the page, and Muse Spark 1.2 is now priced at $1.25 per million input tokens, so neither is a free route on 2026-09-28. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The providers behind Space Bunny Free and LongCat 2.5 Preview Free state a zero-retention policy; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, then lists exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free are described as zero-retention with no training use.",
        "clock": "All ten free models are described as available for a limited time, and no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table, covering input and output tokens and, where a cached read price is listed, cached reads as well. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-29: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model called through the System One endpoint rather than a text generator. Union Alpha Free left the free table by 2026-09-29. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free are served by providers that follow a zero-retention policy and do not use your data for model training; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA free endpoints for trial use only, and use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free follow a zero-retention policy and do not use your data for model training. Jev prompts and other inputs are not used for training, and data is retained in accordance with TypeSafe AI's privacy policy.",
        "clock": "All ten free models are limited-time promotions, and no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-87f1708d5fe130af3e8f",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page, which syncs prices in real time and shows only available models, usable within platform rate limits. Rate limits for free models are fixed values, while paid model limits rise with the account's monthly usage level, and free model calls appear in the account bill at a cost of 0. Per-model rate limits are shown on each model detail page after login. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-28, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and XingChenAGI/Xing4.0-29B from ChinaTelecom, which the model page dates to 2026-09-17. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn with login by SMS or email, and an API key created in the console. The platform's rate limits documentation states that after verifying your identity you can use all free models."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page (real-time price sync, available models only), usable within platform rate limits. Identity verification is required first: the provider states that after verifying your identity you can use all free models, and that free model calls are free and appear as a cost of 0 in the account bill. Rate limits for free models are fixed values, and per-model rate limits are shown on each model detail page after login. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-29, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B (ChinaTelecom, 29B total parameters with 4B active and 256K native context, listed at \uffe50 input and \uffe50 output and released on 2026-09-17), THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), four BAAI bge embedding and reranker models, six speech models including XingChenAGI/XingChenASR-V3.2-Ultra, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall, and one image model, Kwai-Kolors/Kolors. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Identity verification is also required: the provider states that after verifying your identity you can use all free models."
      }
    },
    {
      "id": "free-program-859f13e897d088c4d4d7",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M for input tokens, output tokens and cached input, against list prices of $0.05, $0.15 and $0.005 per 1M. A new user can create an account and an API key without subscribing to a Token Plan, then call the hosted API at https://apihub.agnes-ai.com/v1 using the chat completions endpoint. Free and default text access is rate-limited to 10 requests per minute, following a 50% cut to free and enterprise text limits that the provider dated September 23, 2026. The FAQ says the core models are free to use indefinitely with no time limit, while the pricing page says current offers may vary by model, account eligibility or promotion stage and that the account bill is the source of truth for actual charges.",
        "models_detail": "On 2026-09-27 the pricing page lists four text models. agnes-2.5-flash and agnes-3.0-flash each show a current price of $0 for cached input, input tokens and output tokens, and the page states that all three billing items are currently free for both models. agnes-2.5-pro is billed at its published Pro-tier prices and agnes-2.5-pro-beta at the separate Beta prices, so neither of those is a free route."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens on 2026-09-29. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free and default text access is rate-limited to an effective 10 requests per minute, after a platform notice dated September 23, 2026 halved the text model rate limits for free and enterprise users. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "On 2026-09-29 the pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 input, $0 output and $0 cached input, and states that all three billing items are currently free for both models. agnes-2.5-pro appears on the same table at its paid Pro-tier prices. agnes-2.0-flash no longer appears on the pricing page."
      }
    },
    {
      "id": "free-program-83bd320077add69c46eb",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Mancer's model table lists MythoLite with an input token cost and an output token cost of FREE as of 2026-09-24. Mancer's pricing page states that some models are free while most carry a per token credit cost, and it invites users to test the free models before making a purchase. Free models stay usable even when an account's credit total is negative, so buying credits is not needed to use them.",
        "models_detail": "MythoLite is the only model marked FREE for both input and output tokens on Mancer's model table as of 2026-09-24. It is listed as a LLaMA 2 13B model tagged Creative and Roleplay. The model table keeps each model's context window and maximum output behind a details link and does not show those numbers for MythoLite on the list view."
      },
      "after": {
        "free_terms": "Mancer's model table shows MythoLite with an input and output token cost of FREE on 2026-09-29, so it can be used without buying credits. Mancer's pricing FAQ says some models are free while most have a per token credit cost, and it does not publish the full free list, so MythoLite is the only model the table shows as free. Of the other listed models, only MythoMax carries a visible price on Mancer's pages: 0.14 credits per input token and 0.24 credits per output token. Free models remain usable even when an account's credit balance is negative.",
        "models_detail": "MythoLite is a LLaMA 2 13B model tagged for creative and roleplay use, and on 2026-09-29 it is the one model on Mancer's table shown with a FREE input and output token cost. The table's context and limits columns carry no values for it, and Mancer does not state its context window or maximum output length, so those caps are not stated here."
      }
    },
    {
      "id": "free-program-382b74e1b6e94763a658",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of $1 or more permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset every day. Models that carry a higher request weight have lower caps, published as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute on the Gemini image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page on 2026-09-28, a page the provider marks as updated on 2026-09-27. Nearly all are text models, plus two image models (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, which are available immediately after signing up with email or OAuth. Reaching the permanent daily quotas requires a one-time top-up of $1 or more, so that route needs a payment.",
        "card_required": "yes"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset every day. Models that carry a higher request weight have lower request caps, as low as 10 per day and 1 per minute, and the free catalog table lists the cap for each model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models were listed at $0 input and $0 output on the provider's free catalog page on 2026-09-29, from 16 model authors. Nearly all are text models, plus two that handle images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. Unlocking the daily quotas requires a one-time top-up of any amount from $1, and the provider does not state which payment methods it accepts for that top-up. Free models bill $0 within the quotas, so free usage does not draw the topped-up balance down.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-151e46db9f882f73d35b",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-28) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee, and the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier on Kenari's homepage on 2026-09-28 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog). Kenari's billing documentation says the current quota, rate-limit, and top-up threshold values live on the pricing page because operators can change them at any time."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-29) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per Kenari's homepage Solo plan card on 2026-09-29 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      }
    },
    {
      "id": "free-program-12c139889fdb10b6b70b",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-28: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and points to each model detail page for the retention and training policy of the provider serving that model, so treatment varies by route. Check the underlying provider's policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-29: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation's free-models table is shorter and lists nine of them, leaving out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card is required. The pricing page lists \"No credit card required\" among the Free plan's terms.",
        "data_treatment": "Requesty says it never uses your data for training, and that each model detail page shows the data retention and training policy of the provider serving that model. Those upstream policies differ, so check the model detail page and the underlying provider's policy before sensitive use."
      }
    },
    {
      "id": "free-program-063877610eb79c8cec57",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_record_corrected",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "data_treatment"
      ],
      "before": {
        "data_treatment": "Hetzner's Inference API FAQ says it keeps only the data needed to track usage and, in the future, bill usage, naming request timestamps and token counts. The same FAQ says, 'We do not store the content of request and response and we do not plan on doing so in the future (unless some law requires us to do so).' Hetzner's engineering blog states, 'We do not log any prompt data.'"
      },
      "after": {
        "data_treatment": "Hetzner's Inference API documentation states that the service keeps only what it needs for usage accounting: \"We do only keep data that is necessary in order to track usage and (in the future) bill usage: request timestamps, token counts etc.\" The same page states, \"We do not store the content of request and response and we do not plan on doing so in the future (unless some law requires us to do so).\" Hetzner's engineering blog adds, \"We do not log any prompt data.\""
      }
    },
    {
      "id": "free-program-02a6c0dc826de16bdc69",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-29T00:00:00+00:00",
      "date": "2026-09-29",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment",
        "training_use"
      ],
      "before": {
        "models_detail": "Eight catalog entries were listed at $0.00/1M on 2026-09-28: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it; the plan page itself says nothing more specific about free-inference prompts.",
        "training_use": "yes"
      },
      "after": {
        "models_detail": "Nine catalog entries were listed at $0.00 per 1M on 2026-09-29: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Meituan LongCat 2.5 Preview; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Several of these models also appear separately in the catalog at paid rates. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M.",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it. The same terms add that if you enable Privacy Mode, Nous Research will not store your inference payloads or use them for training, product improvement or support purposes, except as described in the privacy policy. The plan page itself says nothing more specific about free-inference prompts.",
        "training_use": "opt_out"
      }
    },
    {
      "id": "free-program-fbded1e3d8f0c27a9578",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-23. Solar Pro 2 is no longer listed among the free models.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved organizations have access from the date of approval until March 31, 2027, for up to one year."
      },
      "after": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-28. Solar Pro 2 is no longer named in the program.",
        "clock": "The program runs through 2027-03-31 23:00 KST. Approved organizations have access from the date their application is approved until 2027-03-31, for up to one year."
      }
    },
    {
      "id": "free-program-f5058692ff8acc1e04e8",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "clock"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls that work across the whole free catalog, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account to permanent daily quotas: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset daily. Models that carry a higher request weight have lower caps, published per model in the catalog as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the image model gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "On 2026-09-26 the provider's free catalog page listed 60 models at $0 input and $0 output, from 16 model authors. Almost all of them return text, and two return images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the recurring daily quotas requires a one-time top-up of $1 or more, which is a payment rather than a card kept on file, and free usage afterward is billed at $0 instead of being drawn from that balance.",
        "clock": "Ten trial calls per new account that never expire, and permanent daily quotas after a one-time top-up of $1 or more, reset every day with no published end date."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of $1 or more permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset every day. Models that carry a higher request weight have lower caps, published as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute on the Gemini image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page on 2026-09-28, a page the provider marks as updated on 2026-09-27. Nearly all are text models, plus two image models (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, which are available immediately after signing up with email or OAuth. Reaching the permanent daily quotas requires a one-time top-up of $1 or more, so that route needs a payment.",
        "clock": "The ten trial calls given to each new account never expire. After a one-time top-up of $1 or more the daily quotas are permanent and reset every day, and the provider publishes no end date."
      }
    },
    {
      "id": "free-program-e3677b3d62aff03b075d",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. On 2026-09-25 the catalogue lists 20 models, and OVHcloud's product page describes more than 20 models. The available series are NVIDIA Riva, Whisper, bge, gpt-oss, llama, qwen, qwen3guard and stable-diffusion, covering text, reasoning, vision, embedding, image-generation and speech models. Seven catalogue entries are priced Free outright on 2026-09-25: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices nvr-tts-de-de, nvr-tts-es-es, nvr-tts-it-it and nvr-tts-en-us. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR in and 2.7 EUR out per 1M tokens. Individual models can be retired under OVHcloud's lifecycle process. On 2026-09-25 four models are inside their grace period, with decommission dates of 2026-10-01 and 2026-11-04."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-28 the catalog listed 20 models drawn from the Qwen, Qwen3Guard, gpt-oss, Llama, Whisper, bge, Stable Diffusion, and NVIDIA Riva series, and OVHcloud describes the range as more than 20 models. Seven of those entries are priced at Free outright: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the NVIDIA Riva voices nvr-tts-de-de, nvr-tts-en-us, nvr-tts-es-es, and nvr-tts-it-it. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens. Individual models can be retired under OVHcloud's lifecycle process. On 2026-09-28 Mistral-7B-Instruct-v0.3 and Qwen3-Coder-30B-A3B-Instruct-FP8 were listed for decommissioning on 2026-10-01, and Mistral-Nemo-Instruct-2407 and Mistral-Small-3.2-24B-Instruct-2506 on 2026-11-04."
      }
    },
    {
      "id": "free-program-bdb08556a57638fd7359",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "card",
        "data_treatment"
      ],
      "before": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority.",
        "data_treatment": "Requests are processed by volunteer-operated workers; do not treat prompts or outputs as private or suitable for sensitive data."
      },
      "after": {
        "card": "No credit card is required at any point. AI Horde states that everyone can use the service for free, even anonymously, with no credit card and no paywall.",
        "data_treatment": "Requests are processed by volunteer-operated workers around the world. The AI Horde Terms and Conditions and Privacy policy govern the service, including the OpenAI-compatible endpoint at oai.aihorde.net, and the public pages do not state how prompts or outputs are retained or used."
      }
    },
    {
      "id": "free-program-bc195148c6a4bbd722ad",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-27: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Eight catalog entries were listed at $0.00/1M on 2026-09-28: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-b79566bc78a4e197891b",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Amazon states in the Nova Act user guide that API key authentication is free of charge with daily limits, and that the key comes from nova.amazon.com/act and is tied to an Amazon.com account. The separate AWS IAM authentication route is billed to an AWS account and is not part of this free route. The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language: API keys are issued from the developer surface (Terms 1.5), and usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3). The Build with Nova schedule states that Build with Nova is intended as an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). No numeric limits are published publicly."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). The Amazon Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is free of charge with daily limits, while the alternative AWS IAM authentication path bills usage to an AWS account. No numeric limits are published publicly."
      }
    },
    {
      "id": "free-program-88980999f7632ac997d0",
      "program_id": "catalog2-unorouter",
      "event_type": "program_record_corrected",
      "program_name": "UnoRouter free models",
      "provider_label": "UnoRouter",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://unorouter.com/en/docs/platform/models-and-pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "UnoRouter offers models priced at $0 for input and $0 for output, marked with a :free suffix. A :free model routes only to free providers and never uses an account balance, and the same name without the suffix is the paid version. Using the API requires a UnoRouter account and an API key. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, and 503 when every provider for a model is busy. There is no announced expiry, but availability is best effort and the model set shifts over time.",
        "models_detail": "Free models carry a :free suffix, for example gpt-oss-120b:free, and the paid version is the same model name without the suffix. Free models were listed in UnoRouter's own catalog on 2026-09-18, and the set changes as shared pools drain or upstream providers change. Not all free models support tool calls or vision, so check the capability badges on the model page.",
        "signup": "A UnoRouter account and an API key are required. The sign-in methods offered and any payment method requirement at signup are not stated on UnoRouter's models or documentation pages.",
        "clock": "Standing best-effort catalog with a cap of 1 request per minute, per free model, per user, plus upstream provider caps, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit. No published expiry."
      },
      "after": {
        "free_terms": "UnoRouter offers free model routes marked with a :free suffix and priced at $0 input and $0 output. A :free model routes only to free providers and never uses an account balance, while the same model name without the suffix is the paid version billed per token. Calling a free model requires a UnoRouter account and an API key. UnoRouter caps free use at 1 request per minute, per free model, per user, and further limits apply: the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429 with a Retry-After header, and 503 when every free provider for that model is busy. UnoRouter publishes no expiry for the free routes, but availability is best effort and a model leaves the catalog when every channel is out.",
        "models_detail": "Free routes carry a :free suffix in UnoRouter's own catalog, and the set changes as free pools drain or upstream providers change. On 2026-09-28 the newest catalog listings included space-bunny-alpha:free, mimo-v2.6-flash:free, mimo-v2.6-pro:free, axon-1.8-pro:free, axon-1.8-flash:free, axon-1.8-lightning:free, atria-dawn-preview:free, ling-3.0-flash-vl:free, nex-n2.5-mini:free and k2-horizon:free. Not all free models support tool calls or vision, so check the capability badges on the model page.",
        "signup": "A UnoRouter account and an API key are required, and UnoRouter's Quickstart covers getting a key. UnoRouter does not state which account sign-in methods it accepts, and it does not state whether a credit card is required.",
        "clock": "Standing best-effort catalog with no published expiry. Free use is capped at 1 request per minute, per free model, per user, layered with the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit."
      }
    },
    {
      "id": "free-program-402916545f8fefc595e4",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Ten free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table, covering input and output tokens, and cached reads on the rows that list a cached read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-27: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's Jev structured decision model, selected by using the jev-1.13-free model id, and it lists Free input and output with no cached read price. Union Alpha Free is no longer listed as of 2026-09-27. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free are served by providers that follow a zero-retention policy and do not use customer data for model training; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with listed exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during their free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the page says not to submit personal or confidential data. Muse Spark 1.3 Contributor Free trades heavily discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free are covered by the zero-retention policy and are not used for model training."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices ten models at Free for input and output tokens, and nine of those also show Free for cached reads. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-28: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's structured decision model and is called at the System One endpoint. Union Alpha Free no longer appears on the page, and Muse Spark 1.2 is now priced at $1.25 per million input tokens, so neither is a free route on 2026-09-28. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The providers behind Space Bunny Free and LongCat 2.5 Preview Free state a zero-retention policy; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, then lists exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and personal or confidential data should not be submitted. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free are described as zero-retention with no training use."
      }
    },
    {
      "id": "free-program-2c1bc90803449d20e486",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "Free ('\u514d\u8d39') chat models visible on the CN pricing page on 2026-09-27: tencent/Hunyuan-MT-7B (translation), ChinaTelecom's XingChenAGI/Xing4.0-29B (256K context, priced at \u00a50 input and \u00a50 output on the models page, released on 2026-09-17), and PaddlePaddle/PaddleOCR-VL-1.5 (document OCR). Listed as free on the same page on 2026-09-04 and not shown in the collapsed vendor rows read on 2026-09-27, which hide 2 Z-ai models, 6 deepseek-ai models and 30 Qwen models: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and deepseek-ai/DeepSeek-OCR. Also free on 2026-09-27: the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 embedding and reranking models, the XingChenAGI GSR and ASR speech models, Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). The platform documentation states that identity verification must be completed before all free models can be used."
      },
      "after": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-28, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and XingChenAGI/Xing4.0-29B from ChinaTelecom, which the model page dates to 2026-09-17. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn with login by SMS or email, and an API key created in the console. The platform's rate limits documentation states that after verifying your identity you can use all free models."
      }
    },
    {
      "id": "free-program-2b446226200e5a48018e",
      "program_id": "catalog2-zenmux",
      "event_type": "program_record_corrected",
      "program_name": "ZenMux GLM Flash free routes",
      "provider_label": "ZenMux",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://zenmux.ai/z-ai/glm-4.6v-flash-free",
      "changed_fields": [
        "free_terms",
        "signup"
      ],
      "before": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active $0 routes, each marked Active, Free and Rate Limit on 2026-09-23. Access requires a ZenMux account and PAYG API key, and the public pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance.",
        "signup": "A ZenMux account and an API key are required, and ZenMux accepts email, GitHub, or Google sign in. The account balance must remain non-negative for all models including free tiers, and some models require a positive balance. The public pages do not state a card requirement or a positive funded balance for these active free routes."
      },
      "after": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active routes with a $0 effective price on both input and output, each marked Free and Rate Limit on 2026-09-28. Access requires a ZenMux account and an API key, created under either the Pay As You Go plan or the paid Builder Plan subscription. The public pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance. Prior free routes have sunset without becoming permanent, so the lineup is dynamic.",
        "signup": "A ZenMux account and an API key are required. The Quick Start guide has a user sign in with an email address, a GitHub account or a Google account, then choose between Pay As You Go and the Builder Plan subscription, which starts at $20 per month, and then create a key from the matching management page. Account balance must remain non-negative, and the reviewed pages do not prove that a card or a positive funded balance is required for these active free routes."
      }
    },
    {
      "id": "free-program-1d2c5eadf6ba740a7b08",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-27: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Ling 3.0 Tiny, Muse Glimmer 30B, and Nemotron 3.5 Lightning. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan requires no credit card.",
        "data_treatment": "Requesty states that it never uses customer data for training, and that each model detail page shows the data retention and training policy of the provider serving that model. Because routes can involve different model providers, check the serving provider's policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-28: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and points to each model detail page for the retention and training policy of the provider serving that model, so treatment varies by route. Check the underlying provider's policy before sensitive use."
      }
    },
    {
      "id": "free-program-0a3ba71c2014026d1ed1",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Current 1M/250K group on 2026-09-27: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-09-27: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview in its historical list but marks it shut down, so it is not current access."
      },
      "after": {
        "models_detail": "On 2026-09-28 the 1M/250K group listed gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. On the same date the 10M/2.5M group listed gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The quota is shared across each group. The page also retains gpt-4.5-preview-2025-02-27 in the 1M list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-03350f5c8a8107ef62bf",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-28T00:00:00+00:00",
      "date": "2026-09-28",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-27) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models on the Solo tier per Kenari's homepage plan card on 2026-09-27, down from seventeen on 2026-09-17 (it read eleven on 2026-09-04 and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required for the Solo free lane, which the homepage lists as free with no top up. The quick start states that no overseas credit card and no approval wait are needed, and optional top-up is paid by QRIS from Rp 1,000 or by crypto."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-28) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee, and the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models per the Solo tier on Kenari's homepage on 2026-09-28 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog). Kenari's billing documentation says the current quota, rate-limit, and top-up threshold values live on the pricing page because operators can change them at any time.",
        "card": "No card is required for the Solo free lane: the tier is listed as free with no top up, and Kenari's quickstart says no foreign credit card and no approval wait are needed."
      }
    },
    {
      "id": "free-program-ed6b8af24920a589e4d3",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M cached-input, input and output tokens. A new user can create an account and an API key without subscribing to a Token Plan, then call the hosted API at https://apihub.agnes-ai.com/v1. Free and default text access is rate-limited to an effective 10 requests per minute against an allowed 30, reduced by 50 percent under a platform policy change dated 2026-09-23. The pricing page says current offers can vary by model, account eligibility or promotion stage, and that the account bill is the source of truth for actual charges.",
        "models_detail": "agnes-2.5-flash and agnes-3.0-flash are each listed at $0 cached input, $0 input and $0 output, and the pricing page states that all three billing items are currently free for both models. On 2026-09-24 the text pricing table listed only agnes-2.5-flash, agnes-2.5-pro, agnes-2.5-pro-beta and agnes-3.0-flash, with no agnes-2.0-flash entry and no deprecation note; agnes-2.5-pro and agnes-2.5-pro-beta are billed at their published paid prices."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M for input tokens, output tokens and cached input, against list prices of $0.05, $0.15 and $0.005 per 1M. A new user can create an account and an API key without subscribing to a Token Plan, then call the hosted API at https://apihub.agnes-ai.com/v1 using the chat completions endpoint. Free and default text access is rate-limited to 10 requests per minute, following a 50% cut to free and enterprise text limits that the provider dated September 23, 2026. The FAQ says the core models are free to use indefinitely with no time limit, while the pricing page says current offers may vary by model, account eligibility or promotion stage and that the account bill is the source of truth for actual charges.",
        "models_detail": "On 2026-09-27 the pricing page lists four text models. agnes-2.5-flash and agnes-3.0-flash each show a current price of $0 for cached input, input tokens and output tokens, and the page states that all three billing items are currently free for both models. agnes-2.5-pro is billed at its published Pro-tier prices and agnes-2.5-pro-beta at the separate Beta prices, so neither of those is a free route."
      }
    },
    {
      "id": "free-program-dbac9d62043781a6cae1",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-25 the 1M/250K token group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M token group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "Current 1M/250K group on 2026-09-27: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-09-27: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview in its historical list but marks it shut down, so it is not current access."
      }
    },
    {
      "id": "free-program-b4c04ca133397d924900",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page ('real-time prices, available models only'), usable within platform rate limits. Rate limits for the free models are fixed rather than tiered by spend, and per-model limits are listed on the model pages. Free model calls show as zero cost on the account bill. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Added since then: XingChenAGI/Xing4.0-29B from ChinaTelecom, released on 2026-09-17, priced at zero for both input and output and listed as free on the pricing page on 2026-09-25, with a 256K context window. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, Kwai-Kolors/Kolors image generation, and several ASR and speech models including the XingChenAGI ASR family, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Identity verification is also required before the free models can be used: the platform rate limits documentation states that after verifying your identity you can use all free models."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page, which syncs prices in real time and shows only available models, usable within platform rate limits. Rate limits for free models are fixed values, while paid model limits rise with the account's monthly usage level, and free model calls appear in the account bill at a cost of 0. Per-model rate limits are shown on each model detail page after login. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models visible on the CN pricing page on 2026-09-27: tencent/Hunyuan-MT-7B (translation), ChinaTelecom's XingChenAGI/Xing4.0-29B (256K context, priced at \u00a50 input and \u00a50 output on the models page, released on 2026-09-17), and PaddlePaddle/PaddleOCR-VL-1.5 (document OCR). Listed as free on the same page on 2026-09-04 and not shown in the collapsed vendor rows read on 2026-09-27, which hide 2 Z-ai models, 6 deepseek-ai models and 30 Qwen models: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and deepseek-ai/DeepSeek-OCR. Also free on 2026-09-27: the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 embedding and reranking models, the XingChenAGI GSR and ASR speech models, Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). The platform documentation states that identity verification must be completed before all free models can be used."
      }
    },
    {
      "id": "free-program-af53426e3f3588a6fadf",
      "program_id": "catalog2-meganova",
      "event_type": "program_record_corrected",
      "program_name": "MegaNova free-quota model (Mistral Small 3.2)",
      "provider_label": "Meganova",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://www.meganova.ai/mistralai/Mistral-Small-3.2-24B-Instruct-2506",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "clock"
      ],
      "before": {
        "free_terms": "MegaNova hosts `mistralai/Mistral-Small-3.2-24B-Instruct-2506` as a recurring free-quota text model. Free Tier 1 registration requires no credit card and includes 50 requests per day for this model, within account rate limits of 60 requests per minute and 200,000 tokens per minute. Free quota is granted per account and resets daily at 00:00 UTC. Documentation states that Tier 2 and above keep free-quota access with higher limits, but the published quota table does not list per-tier request numbers for this model. Once the daily quota is exhausted, requests fail with a quota error unless the optional \"Allow paid usage after free quota\" switch is enabled, so this is genuine standing free hosted inference rather than unlimited `$0` usage.",
        "models_detail": "mistralai/Mistral-Small-3.2-24B-Instruct-2506 is the documented recurring free-quota text model, listed as Free on its model page with a context length of 8,192 tokens.",
        "card": "No credit card is required: the Tier 1 page states that registration is free with no credit card required. The optional paid fallback is a separate opt-in setting.",
        "clock": "Standing daily quota: 50 requests/day at Tier 1, resetting at 00:00 UTC; higher verified tiers receive larger daily quotas."
      },
      "after": {
        "free_terms": "MegaNova hosts mistralai/Mistral-Small-3.2-24B-Instruct-2506 as a recurring free-quota text model, listed as free on 2026-09-27. Free Tier 1 registration requires no credit card and includes 50 requests per day for this model, within Tier 1 rate limits of 60 requests per minute and 200,000 tokens per minute. Free quota is granted per account and resets daily at 00:00 UTC. MegaNova documents that Tier 2 and above keep free quota access with higher limits. Once the daily free quota runs out, requests normally fail with a quota error unless the account has turned on the optional paid fallback, so this is genuine standing free hosted inference but not unlimited zero-cost usage.",
        "models_detail": "mistralai/Mistral-Small-3.2-24B-Instruct-2506 is the documented recurring free-quota text model, labeled Free on its model page with a context length of 8,192 tokens on 2026-09-27.",
        "card": "MegaNova states that Tier 1 registration is free and requires no credit card.",
        "clock": "Standing daily quota: 50 requests per day at Tier 1, resetting at 00:00 UTC. MegaNova documents higher limits for Tier 2 and above."
      }
    },
    {
      "id": "free-program-8511af1528ca935e3a42",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-25: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in / $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-27: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-74780ea9169a0770c841",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated 2025-12-02, contain no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the environment does not support deploying production scale Developer Offerings (S-1.1). For Amazon Nova Act, the AWS user guide states that API key authentication from nova.amazon.com is free of charge with daily limits, while AWS IAM authentication is the separate path on which usage charges apply to your AWS account. No numeric quota or rate limit figures are published publicly.",
        "card": "not required; the Nova Terms of Use contain no payment terms, and the Nova Act user guide describes API key authentication as free of charge with daily limits"
      },
      "after": {
        "free_terms": "Amazon states in the Nova Act user guide that API key authentication is free of charge with daily limits, and that the key comes from nova.amazon.com/act and is tied to an Amazon.com account. The separate AWS IAM authentication route is billed to an AWS account and is not part of this free route. The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language: API keys are issued from the developer surface (Terms 1.5), and usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3). The Build with Nova schedule states that Build with Nova is intended as an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). No numeric limits are published publicly.",
        "card": "Not required. The Nova Act user guide states that API key authentication is free of charge with daily limits, and no payment terms appear anywhere in the Nova Terms of Use."
      }
    },
    {
      "id": "free-program-6adbc89b2b86bc0bb64a",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-23: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty says it never uses customer data for training, and it points to each model's detail page for the underlying provider's retention and training policy. Routes can involve different model providers, so check the model detail page and that provider's own policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-27: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Ling 3.0 Tiny, Muse Glimmer 30B, and Nemotron 3.5 Lightning. Requesty says the set is free for now and announces changes in its changelog.",
        "data_treatment": "Requesty states that it never uses customer data for training, and that each model detail page shows the data retention and training policy of the provider serving that model. Because routes can involve different model providers, check the serving provider's policy before sensitive use."
      }
    },
    {
      "id": "free-program-35b309bead61f51d1ac6",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (18 models on 2026-09-24) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Eighteen free models on 2026-09-24 per the Solo tier row on Kenari's homepage (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required. The Solo tier is listed as free with no top up, account creation asks only for an email address and password or a Google sign-in, and the quickstart says no overseas credit card is needed. Top-ups are optional and are paid by QRIS from Rp 1.000 or by crypto."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-27) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen provider-hosted :free text models on the Solo tier per Kenari's homepage plan card on 2026-09-27, down from seventeen on 2026-09-17 (it read eleven on 2026-09-04 and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required for the Solo free lane, which the homepage lists as free with no top up. The quick start states that no overseas credit card and no approval wait are needed, and optional top-up is paid by QRIS from Rp 1,000 or by crypto."
      }
    },
    {
      "id": "free-program-2d7fb48103fc46df15f1",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Ten free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock",
        "training_use"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices nine models at Free for input and output tokens. Eight of those nine also show Free for cached reads, and Jev 1.13 Free lists no cached read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-24: Big Pickle, Space Bunny Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Space Bunny Free, MiMo-V2.6-Flash Free and Jev 1.13 Free joined the free table by 2026-09-24. Union Alpha Free left the free table by 2026-09-24, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free's provider follows a zero-retention policy and does not use your data for model training. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Jev 1.13 Free is a structured decision model reached with the model id jev-1.13-free. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, with exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and Zen says not to submit personal or confidential data to them. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free's provider follows a zero-retention policy and does not use your data for model training.",
        "clock": "All nine free models are described as available for a limited time, and no end date is stated for any of them.",
        "training_use": "yes"
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices ten models at Free on its pricing table, covering input and output tokens, and cached reads on the rows that list a cached read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-27: Big Pickle, Space Bunny Free, LongCat 2.5 Preview Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of Zen's Jev structured decision model, selected by using the jev-1.13-free model id, and it lists Free input and output with no cached read price. Union Alpha Free is no longer listed as of 2026-09-27. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free and LongCat 2.5 Preview Free are served by providers that follow a zero-retention policy and do not use customer data for model training; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with listed exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may have collected data used to improve those models during their free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the page says not to submit personal or confidential data. Muse Spark 1.3 Contributor Free trades heavily discounted token pricing for permission to use prompts and completions to train future Meta models. Space Bunny Free and LongCat 2.5 Preview Free are covered by the zero-retention policy and are not used for model training.",
        "clock": "All ten free models are described as available for a limited time, and no end date is stated for any of them.",
        "training_use": "varies"
      }
    },
    {
      "id": "free-program-0eeafbb16e93ec477cbb",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-27T00:00:00+00:00",
      "date": "2026-09-27",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "clock"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free catalog and reset daily. Some models carry lower caps because one call to them counts as several requests: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, or 10 per day and 1 per minute. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page (page updated 2026-09-24), nearly all text models plus two that return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required to create an account or to use the 10 trial calls. Reaching the permanent daily quotas requires one payment, a single top-up of $1 or more, which free usage does not draw down because the free models bill $0 for input and output.",
        "clock": "The 10 trial calls never expire. After a one-time top-up of $1 or more, the daily quotas are permanent and reset every day, and the provider publishes no end date."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls that work across the whole free catalog, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account to permanent daily quotas: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset daily. Models that carry a higher request weight have lower caps, published per model in the catalog as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the image model gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "On 2026-09-26 the provider's free catalog page listed 60 models at $0 input and $0 output, from 16 model authors. Almost all of them return text, and two return images: gpt-image-2-free and gemini-3.1-flash-image-preview-free. The roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the recurring daily quotas requires a one-time top-up of $1 or more, which is a payment rather than a card kept on file, and free usage afterward is billed at $0 instead of being drawn from that balance.",
        "clock": "Ten trial calls per new account that never expire, and permanent daily quotas after a one-time top-up of $1 or more, reset every day with no published end date."
      }
    },
    {
      "id": "free-event-e2018f261e59ee6d21af",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nex-agi/nex-n2.5-mini:free",
      "model": "Nex AGI: Nex-N2.5-Mini",
      "model_key": "nexn25minifree",
      "detected_at": "2026-09-26T22:54:26+00:00",
      "date": "2026-09-26",
      "evidence_url": "https://openrouter.ai/nex-agi/nex-n2.5-mini:free"
    },
    {
      "id": "free-event-9b04c13dd832e3d0b0e8",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nex-agi/nex-n2.5-pro:free",
      "model": "Nex AGI: Nex-N2.5-Pro",
      "model_key": "nexn25profree",
      "detected_at": "2026-09-26T22:54:26+00:00",
      "date": "2026-09-26",
      "evidence_url": "https://openrouter.ai/nex-agi/nex-n2.5-pro:free"
    },
    {
      "id": "free-event-3ef57970ac92fd5927e2",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "z-ai/glm-5.2:free",
      "model": "Z.ai: GLM 5.2",
      "model_key": "glm52free",
      "detected_at": "2026-09-26T22:54:26+00:00",
      "date": "2026-09-26",
      "evidence_url": "https://openrouter.ai/z-ai/glm-5.2:free"
    },
    {
      "id": "free-program-f31c60a747335f3b1bcd",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "signup",
        "card"
      ],
      "before": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated on 2025-12-02, are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), and usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3). The Build with Nova schedule states that Build with Nova is intended as an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). The Amazon Nova Act user guide lists API key authentication as free of charge with daily limits, and lists AWS IAM authentication separately as a route where usage charges apply to your AWS account. No numeric quotas or daily limit figures are published publicly.",
        "signup": "Required: the API key comes from nova.amazon.com/act and is tied to an Amazon.com account.",
        "card": "Not required. The Nova Terms of Use contain no payment or fee terms, and the Amazon Nova Act user guide lists API key authentication as free of charge with daily limits, in contrast to AWS IAM authentication, where usage charges apply to an AWS account."
      },
      "after": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated 2025-12-02, contain no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states that the environment does not support deploying production scale Developer Offerings (S-1.1). For Amazon Nova Act, the AWS user guide states that API key authentication from nova.amazon.com is free of charge with daily limits, while AWS IAM authentication is the separate path on which usage charges apply to your AWS account. No numeric quota or rate limit figures are published publicly.",
        "signup": "required; Amazon.com consumer account sign-in, then an API key from the developer surface. The Nova Act user guide states that the API key is tied to your Amazon.com account.",
        "card": "not required; the Nova Terms of Use contain no payment terms, and the Nova Act user guide describes API key authentication as free of charge with daily limits"
      }
    },
    {
      "id": "free-program-d9d1aac63b42564ee74a",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the AI Endpoints catalogue. The catalogue listed 20 models on 2026-09-24, and the product page describes a constantly updated range of more than 20 popular, open-weight models. The listed models include Llama, Qwen, GPT-OSS, Whisper speech to text, NVIDIA Riva text to speech, bge and Qwen3 embeddings, and Stable Diffusion XL image generation. Individual models can be retired under OVHcloud's lifecycle process, and a model whose retirement has been announced stays reachable through the API during its grace period while being hidden from the website catalogue. The authenticated catalogue also prices some entries at Free outright: on 2026-09-24 those were the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for US English, German, Spanish and Italian. Qwen3.8-27B itself is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. On 2026-09-25 the catalogue lists 20 models, and OVHcloud's product page describes more than 20 models. The available series are NVIDIA Riva, Whisper, bge, gpt-oss, llama, qwen, qwen3guard and stable-diffusion, covering text, reasoning, vision, embedding, image-generation and speech models. Seven catalogue entries are priced Free outright on 2026-09-25: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices nvr-tts-de-de, nvr-tts-es-es, nvr-tts-it-it and nvr-tts-en-us. Other entries are paid, for example Qwen3.8-27B at 0.4 EUR in and 2.7 EUR out per 1M tokens. Individual models can be retired under OVHcloud's lifecycle process. On 2026-09-25 four models are inside their grace period, with decommission dates of 2026-10-01 and 2026-11-04."
      }
    },
    {
      "id": "free-program-c4c221762c1b22036771",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-22, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and ChinaTelecom's XingChenAGI/Xing4.0-29B, which the model page lists at \u00a50 input and \u00a50 output. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, plus identity verification before the free models can be used, with the API key created in the console (SMS login favors Chinese numbers, email login supported)"
      },
      "after": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Added since then: XingChenAGI/Xing4.0-29B from ChinaTelecom, released on 2026-09-17, priced at zero for both input and output and listed as free on the pricing page on 2026-09-25, with a 256K context window. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, Kwai-Kolors/Kolors image generation, and several ASR and speech models including the XingChenAGI ASR family, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported). Identity verification is also required before the free models can be used: the platform rate limits documentation states that after verifying your identity you can use all free models."
      }
    },
    {
      "id": "free-program-8d59a7cc2686551c65a4",
      "program_id": "together-ternary-bonsai-free",
      "event_type": "program_record_corrected",
      "program_name": "Prism-ML/Ternary-Bonsai-27B",
      "provider_label": "Together",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://www.together.ai/pricing"
    },
    {
      "id": "free-program-6b2ec2f9a6e37cb69362",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-24: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-25: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in / $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-6af796e51bafd49cccbb",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Current 1M/250K group on 2026-09-24: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-09-24: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview in its historical list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      },
      "after": {
        "models_detail": "On 2026-09-25 the 1M/250K token group lists gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. The 10M/2.5M token group lists gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview-2025-02-27 in the 1M group but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-5fe7b06da0d1cb8cc1f9",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-24 the pricing page marks standard Free tier input and output as free of charge for Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini 2.5 Flash-Lite Preview, Gemini Embedding, and Gemini Robotics ER 1.6 Preview no longer carry pricing tables on that page. This is model-specific: other listed models, including Gemini 3.1 Pro Preview and the image, video, and music models, along with several batch and flex modes, show Free tier 'Not available.'"
      },
      "after": {
        "models_detail": "On the Gemini API pricing page as of 2026-09-25, the models whose Standard tables mark Free tier input and output as free of charge are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live for streaming transcription, Gemini 3.5 Transcribe for diarizing transcription, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview are no longer listed on the pricing page. This is model specific: other listed models, including the Omni Flash, Pro, image, video, and music entries, show Free tier as Not available, and the batch, flex, and priority modes differ by model."
      }
    },
    {
      "id": "free-program-58cad57ae9b15cb5534d",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across the whole free catalog. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free models: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across the catalog and reset every day. Models that carry a higher request weight have lower caps: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free (on 2026-09-24). Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page (page updated 2026-09-23, read 2026-09-24), nearly all returning text, plus two image-capable models (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, shared across the free catalog and reset daily. Some models carry lower caps because one call to them counts as several requests: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, or 10 per day and 1 per minute. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page (page updated 2026-09-24), nearly all text models plus two that return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time."
      }
    },
    {
      "id": "free-program-2937c5663b2c1f094ed7",
      "program_id": "morph-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Morph free tier (200 requests per month)",
      "provider_label": "Morph",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://www.morphllm.com/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "clock"
      ],
      "before": {
        "free_terms": "Morph documents a recurring $0 free tier of 200 API requests per month. Calls to its OpenAI-compatible hosted general LLMs count against that request allowance; after the allowance is exhausted, paid per-token or plan-credit pricing applies.",
        "models_detail": "200 hosted LLM API requests per month",
        "clock": "Recurring monthly allowance of 200 requests; no end date published."
      },
      "after": {
        "free_terms": "Morph documents a recurring $0 free tier of 200 API requests per month. On 2026-09-25 the pricing page counts one request for each API call to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, or Rerank against that 200 per month cap, and states that per-token or plan-credit pricing applies once the cap is exhausted. The same page lists a Free subscription plan at $0 per month carrying 250K credits and low rate limits, with 1 credit worth about $0.00001 and credits applying to Fast Apply, WarpGrep, Compact, Reflex, Embeddings, and Rerank. Morph's hosted chat models are listed with per-token rates, and the page does not state that the free request allowance or the free credits cover them.",
        "models_detail": "On 2026-09-25 the documented free allowance covers Fast Apply, WarpGrep, Compact, Reflex, Embeddings, and Rerank: 200 requests per month, plus 250K credits per month on the $0 Free plan. The hosted chat models listed on the pricing page that day, Qwen 3.8, GLM-5.3, GLM-5.3-Flash, and DeepSeek V4.1 Flash, carry per-token prices and are not named in the free-tier request or credit list.",
        "clock": "Recurring monthly allowance: 200 requests per month, plus 250K credits on the $0 Free plan that reset monthly on the subscription date and do not roll over. No end date published."
      }
    },
    {
      "id": "free-program-009fd75a4c223aa063ad",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-09-25T00:00:00+00:00",
      "date": "2026-09-25",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not supported on the Lite plan. Quota limits reset on the first of every month, and reaching any quota limit suspends the service for that month."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan."
      }
    },
    {
      "id": "free-event-7ec9f624e8b2356ec243",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "inclusionai/ling-3.0-flash-vl:free",
      "model": "inclusionAI: Ling 3.0 Flash VL",
      "model_key": "ling30flashvlfree",
      "detected_at": "2026-09-24T11:30:15+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-vl:free"
    },
    {
      "id": "free-program-ed1a9b3ca6ad8c0ef4d7",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). The Amazon Nova Act user guide, read on 2026-09-20, states that API key authentication using a key from nova.amazon.com/act tied to an Amazon.com account is free of charge with daily limits, while a separate AWS IAM authentication route carries usage charges on an AWS account. No numeric limits are published publicly.",
        "card": "Not required. The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and no payment terms appear anywhere in the Nova Terms of Use. Charges apply only on the separate AWS IAM route, which is not this free offering."
      },
      "after": {
        "free_terms": "The Nova.amazon.com Terms of Use, last updated on 2025-12-02, are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), and usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3). The Build with Nova schedule states that Build with Nova is intended as an environment for experimentation and does not support deploying production scale Developer Offerings (S-1.1). The Amazon Nova Act user guide lists API key authentication as free of charge with daily limits, and lists AWS IAM authentication separately as a route where usage charges apply to your AWS account. No numeric quotas or daily limit figures are published publicly.",
        "card": "Not required. The Nova Terms of Use contain no payment or fee terms, and the Amazon Nova Act user guide lists API key authentication as free of charge with daily limits, in contrast to AWS IAM authentication, where usage charges apply to an AWS account."
      }
    },
    {
      "id": "free-program-ea7e267b8cd18657dcc7",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Nine free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices eight models at Free for input and output tokens, and seven of those eight also show a Free cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-23: Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured-decision model reached at the systemone endpoint, and it is the one free row with no cached-read price listed. Union Alpha Free left the free table by 2026-09-23; Muse Spark 1.2 Contributor Free left by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints whose use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models.",
        "clock": "All eight are limited-time promotions; no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices nine models at Free for input and output tokens. Eight of those nine also show Free for cached reads, and Jev 1.13 Free lists no cached read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-24: Big Pickle, Space Bunny Free, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Space Bunny Free, MiMo-V2.6-Flash Free and Jev 1.13 Free joined the free table by 2026-09-24. Union Alpha Free left the free table by 2026-09-24, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Space Bunny Free's provider follows a zero-retention policy and does not use your data for model training. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Jev 1.13 Free is a structured decision model reached with the model id jev-1.13-free. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use your data for model training, with exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and Zen says not to submit personal or confidential data to them. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Space Bunny Free's provider follows a zero-retention policy and does not use your data for model training.",
        "clock": "All nine free models are described as available for a limited time, and no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-e2d29cd12d32139d724c",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier that needs no top up and no card, with 50 requests per day across the current free-model roster (17 models on 2026-09-23) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Kenari's billing documentation states that the quota, per-minute rate, and top-up threshold in force at any moment are published on the pricing page and can be reset by the operator at any time, so confirm the current figures there before relying on them. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Seventeen provider-hosted :free text models per Kenari's homepage counter on 2026-09-23 (it also read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (18 models on 2026-09-24) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Eighteen free models on 2026-09-24 per the Solo tier row on Kenari's homepage (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog)."
      }
    },
    {
      "id": "free-program-a2ba4ab450e8bb48abb8",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. On 2026-09-23 the catalogue listed 20 models spanning embeddings, image generation, LLM guard, large language models, reasoning LLMs, speech to text, text to speech, and visual LLMs, drawn from the Llama, Qwen, GPT-OSS, Whisper, bge, NVIDIA Riva, and Stable Diffusion families. Individual models can be retired under OVHcloud's lifecycle process. Seven catalogue entries are priced at Free outright on 2026-09-23: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the nvr-tts-de-de, nvr-tts-es-es, nvr-tts-it-it, and nvr-tts-en-us text-to-speech voices. Qwen3.8-27B is paid at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the AI Endpoints catalogue. The catalogue listed 20 models on 2026-09-24, and the product page describes a constantly updated range of more than 20 popular, open-weight models. The listed models include Llama, Qwen, GPT-OSS, Whisper speech to text, NVIDIA Riva text to speech, bge and Qwen3 embeddings, and Stable Diffusion XL image generation. Individual models can be retired under OVHcloud's lifecycle process, and a model whose retirement has been announced stays reachable through the API during its grace period while being hidden from the website catalogue. The authenticated catalogue also prices some entries at Free outright: on 2026-09-24 those were the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for US English, German, Spanish and Italian. Qwen3.8-27B itself is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      }
    },
    {
      "id": "free-program-75a8522ce5deb04985eb",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-23 the models whose pricing tables mark standard Free tier input and output as free of charge are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries, Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription, both marked free of charge on 2026-09-04. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini Robotics ER 1.6 Preview, and a separate Gemini 2.5 Flash-Lite Preview entry no longer appear on the pricing page. This is model-specific: other listed models, including the Pro preview, image, video, and music models, show Free tier 'Not available', and discounted or batch modes may also say 'Not available' for models whose standard mode is free."
      },
      "after": {
        "models_detail": "On 2026-09-24 the pricing page marks standard Free tier input and output as free of charge for Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini 2.5 Flash-Lite Preview, Gemini Embedding, and Gemini Robotics ER 1.6 Preview no longer carry pricing tables on that page. This is model-specific: other listed models, including Gemini 3.1 Pro Preview and the image, video, and music models, along with several batch and flex modes, show Free tier 'Not available.'"
      }
    },
    {
      "id": "free-program-72fd361865340acf7c02",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset daily. Some models carry lower request caps, down to 10 per day and 1 per minute, because one call on those models counts as several requests.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page on 2026-09-23, nearly all text models plus two that return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The daily free quotas require a one-time top-up of at least $1, and the provider does not state which payment methods that top-up accepts.",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across the whole free catalog. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free models: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across the catalog and reset every day. Models that carry a higher request weight have lower caps: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free (on 2026-09-24). Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models from 16 model authors listed at $0 input and $0 output on the provider's free catalog page (page updated 2026-09-23, read 2026-09-24), nearly all returning text, plus two image-capable models (gpt-image-2-free and gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required to create an account or to use the 10 trial calls. Reaching the permanent daily quotas requires one payment, a single top-up of $1 or more, which free usage does not draw down because the free models bill $0 for input and output.",
        "card_required": "yes"
      }
    },
    {
      "id": "free-program-3cf1e4e6141213c1b282",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens; agnes-2.0-flash is deprecated and kept on the page as a historical record. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to about 20 RPM. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "agnes-2.5-flash and agnes-3.0-flash are listed at $0 input, $0 output and $0 cached input (the pricing page says all three are currently free for both models); agnes-2.0-flash is marked deprecated, with its $0 prices kept only as historical records."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M cached-input, input and output tokens. A new user can create an account and an API key without subscribing to a Token Plan, then call the hosted API at https://apihub.agnes-ai.com/v1. Free and default text access is rate-limited to an effective 10 requests per minute against an allowed 30, reduced by 50 percent under a platform policy change dated 2026-09-23. The pricing page says current offers can vary by model, account eligibility or promotion stage, and that the account bill is the source of truth for actual charges.",
        "models_detail": "agnes-2.5-flash and agnes-3.0-flash are each listed at $0 cached input, $0 input and $0 output, and the pricing page states that all three billing items are currently free for both models. On 2026-09-24 the text pricing table listed only agnes-2.5-flash, agnes-2.5-pro, agnes-2.5-pro-beta and agnes-3.0-flash, with no agnes-2.0-flash entry and no deprecation note; agnes-2.5-pro and agnes-2.5-pro-beta are billed at their published paid prices."
      }
    },
    {
      "id": "free-program-3c61437a4c791888f26e",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Current 1M/250K group: gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview in its historical list but marks it shut down, so it is not current access."
      },
      "after": {
        "models_detail": "Current 1M/250K group on 2026-09-24: gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gpt-5.2-2025-12-11, gpt-5.1-2025-11-13, gpt-5.1-codex, gpt-5-codex, gpt-5-2025-08-07, gpt-5-chat-latest, gpt-4.1-2025-04-14, gpt-4o-2024-05-13, gpt-4o-2024-08-06, gpt-4o-2024-11-20, o3-2025-04-16, o1-preview-2024-09-12, and o1-2024-12-17. Current 10M/2.5M group on 2026-09-24: gpt-5.6-terra, gpt-5.6-luna, gpt-5.4-mini-2026-03-17, gpt-5.4-nano-2026-03-17, gpt-5.1-codex-mini, gpt-5-mini-2025-08-07, gpt-5-nano-2025-08-07, gpt-4.1-mini-2025-04-14, gpt-4.1-nano-2025-04-14, gpt-4o-mini-2024-07-18, o4-mini-2025-04-16, o1-mini-2024-09-12, and codex-mini-latest. The page also retains gpt-4.5-preview in its historical list but marks it deprecated and shut down as of 7/14/25, so it is not current access."
      }
    },
    {
      "id": "free-program-3b1507fc2e982e5bee99",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-23: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is no longer in the free set and is listed at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Eight catalog entries at $0.00/1M on 2026-09-24: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; Space Bunny Alpha; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-0dc1d26af1bd5e123b54",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan, a free quota based service plan with 300,000 foundation model inference tokens per month, 20 capacity unit hours per month for machine learning workloads, and 100 document text classification and extraction pages per month, with a plan level limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan. Reaching any Lite plan quota suspends the service for the remainder of that month.",
        "models_detail": "Ready to use IBM and third party foundation models are available to the Lite instance, subject to regional and current catalog availability. IBM's supported models page describes both IBM and third party foundation models provided with watsonx.ai for inferencing, and states that foundation model availability varies by data center location. Custom foundation models are not available on the Lite plan. An exact immutable Lite model roster is not stated."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction pages per month, and a plan-level limit of 2 inference requests per second. Custom foundation models are not supported on the Lite plan. Quota limits reset on the first of every month, and reaching any quota limit suspends the service for that month.",
        "models_detail": "Ready-to-use IBM and third-party foundation models are available for per-token inferencing, subject to regional and current catalog availability. On 2026-09-24 the IBM models provided for inferencing were granite-4-h-small, granite-guardian-3-8b, and granite-8b-code-instruct, with the last two marked deprecated, and the third-party models provided for inferencing came from OpenAI, Meta, Mistral AI, and BigScience. IBM does not publish a separate model roster for the Lite plan, and custom foundation models are not supported on it. Deploy-on-demand models are billed by the hour."
      }
    },
    {
      "id": "free-program-0021d933c447fd7ba5a4",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-24T00:00:00+00:00",
      "date": "2026-09-24",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "On 2026-09-21 Mancer's model table lists MythoLite as the only model priced FREE for both input and output tokens. Mancer's pricing FAQ says that some models are free while most have a per token credit cost, and it presents the free models as a way to test the API before making a purchase. Free use does not depend on a paid balance: Mancer states that a user can still use free models with a negative credit count. MythoMax, the other LLaMA 2 13B model on the list, is priced at 0.14 credits per input token and 0.24 credits per output token.",
        "models_detail": "MythoLite only: on 2026-09-21 the model table lists it as a LLaMA 2 13B model tagged Creative, Roleplay, with input and output token cost both shown as FREE. The listing does not state MythoLite's context window or maximum output length.",
        "signup": "required: the provider's sign-up surface says to create a Mancer AI account to start inferencing"
      },
      "after": {
        "free_terms": "Mancer's model table lists MythoLite with an input token cost and an output token cost of FREE as of 2026-09-24. Mancer's pricing page states that some models are free while most carry a per token credit cost, and it invites users to test the free models before making a purchase. Free models stay usable even when an account's credit total is negative, so buying credits is not needed to use them.",
        "models_detail": "MythoLite is the only model marked FREE for both input and output tokens on Mancer's model table as of 2026-09-24. It is listed as a LLaMA 2 13B model tagged Creative and Roleplay. The model table keeps each model's context window and maximum output behind a details link and does not show those numbers for MythoLite on the list view.",
        "signup": "An account is required. Mancer's pricing page notes that the usage dashboard and the billing page both require login, and Mancer provides no interface of its own, so a separate client application is needed to send requests to the API."
      }
    },
    {
      "id": "free-program-dfef60508a9725eac70d",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-22. Solar Pro 2 is not named in the program.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved organizations get access for up to one year, running from approval until 2027-03-31, so an organization approved later in the window gets less than a full year."
      },
      "after": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-23. Solar Pro 2 is no longer listed among the free models.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved organizations have access from the date of approval until March 31, 2027, for up to one year."
      }
    },
    {
      "id": "free-program-c47bcd7b1c76de7d15ba",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls, shared across all free models, with no credit card and no expiry. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across the catalog and reset every day. Individual models can carry lower request caps, stepping down through 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute, because a call to a heavier model counts as several requests (caps read on 2026-09-22). Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page, which the provider marked updated on 2026-09-21: nearly all text models, plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The roster changes over time.",
        "card": "No credit card is required to open an account or to use the 10 trial calls. Moving to the permanent daily quotas requires a one-time top-up of at least $1, which is a payment rather than a card kept on file.",
        "card_required": "yes"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of any amount from $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, all reset daily. Some models carry lower request caps, down to 10 per day and 1 per minute, because one call on those models counts as several requests.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page on 2026-09-23, nearly all text models plus two that return images (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls. The daily free quotas require a one-time top-up of at least $1, and the provider does not state which payment methods that top-up accepts.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-c291cb5b732168e823f8",
      "program_id": "catalog2-zenmux",
      "event_type": "program_record_corrected",
      "program_name": "ZenMux GLM Flash free routes",
      "provider_label": "ZenMux",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://zenmux.ai/z-ai/glm-4.6v-flash-free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "clock"
      ],
      "before": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active $0/$0 routes on 2026-09-20, each marked Free and Rate Limit. Access requires a ZenMux account and an API key created after signing in. ZenMux offers a Pay As You Go plan and a paid Builder subscription, and the public pages do not state which key type reaches these free routes, nor do they state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance. Prior free routes have sunset without becoming permanent, so the lineup is dynamic.",
        "models_detail": "On 2026-09-18 the active free routes are z-ai/glm-4.6v-flash-free, published 2025-12-08, and z-ai/glm-4.7-flash-free, published 2026-01-19. ZenMux changes its free lineup over time, so confirm a route on its own model page before relying on it.",
        "clock": "The two active routes carry no stated expiry date. ZenMux changes its free lineup over time, so a route can be withdrawn without notice."
      },
      "after": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active $0 routes, each marked Active, Free and Rate Limit on 2026-09-23. Access requires a ZenMux account and PAYG API key, and the public pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance.",
        "models_detail": "On 2026-09-23 the free lineup is two routes, both showing Active on their ZenMux model pages: z-ai/glm-4.6v-flash-free, described as the free version of GLM-4.6V, and z-ai/glm-4.7-flash-free, a 30B class model tuned for agentic coding. Each route is served by the Z.ai and BigModel providers at an effective $0 per million tokens for both input and output.",
        "clock": "No fixed expiry is stated for either route. Both model pages showed Active with no end date on 2026-09-23."
      }
    },
    {
      "id": "free-program-c0d8a5f2213f259c6e2d",
      "program_id": "arli-free-plan",
      "event_type": "program_record_corrected",
      "program_name": "Arli AI models (free plan)",
      "provider_label": "Arli AI",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://www.arliai.com/pricing",
      "changed_fields": [
        "free_terms",
        "data_treatment"
      ],
      "before": {
        "free_terms": "Arli AI lists a $0 Free plan with delayed and lower-priority responses, a maximum 12K-token context, one request at a time, no image-generation tools, and the exact quota wording \"5 times / 2 days trial of all models.\" The wording does not clearly state whether this means five uses every two days or a one-time two-day trial.",
        "data_treatment": "Arli operates on a zero-log basis. Its current terms and privacy policy say prompts, inputs, and generated outputs are processed only transiently, immediately discarded, and not stored, retained, or accessed beyond returning the response. This supports no use of prompts or outputs for training."
      },
      "after": {
        "free_terms": "Arli AI lists a $0 Free plan with delayed responses that get slower with more requests, a maximum 12K-token context, one request at a time, no image-generation tools, and the exact quota wording \"5 times / 2 days trial of all models.\" The wording does not clearly state whether this means five uses every two days or a one-time two-day trial.",
        "data_treatment": "Arli describes a zero-log approach for content. Its current terms and privacy policy say prompts, inputs, and generated outputs are processed only transiently, immediately discarded, and not stored, retained, or accessed beyond returning the response. This supports no use of prompts or outputs for training. Arli does log usage metadata, including the number of requests sent, the models used, and the parameters of those requests, and retains that metadata for up to 60 days before automatic deletion. It also automatically collects technical data including the user's IP address for security, abuse prevention, and operation of the services."
      }
    },
    {
      "id": "free-program-ba4b1c44169a4b895efe",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Eight free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices eight models at Free for input and output tokens, and seven of those also show a free cached read. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-22: Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model reached at the systemone endpoint rather than a chat model, and it is the one free row with no cached-read price. MiMo-V2.6-Flash Free joined the free table by 2026-09-22. Union Alpha Free left the free table by 2026-09-22, Muse Spark 1.2 Contributor Free left by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints whose use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Jev 1.13 Free is not named among the exceptions."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices eight models at Free for input and output tokens, and seven of those eight also show a Free cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-23: Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured-decision model reached at the systemone endpoint, and it is the one free row with no cached-read price listed. Union Alpha Free left the free table by 2026-09-23; Muse Spark 1.2 Contributor Free left by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints whose use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models."
      }
    },
    {
      "id": "free-program-acfa631beee6fd2cf53d",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-22: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card is required on the Free plan.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policies of the underlying model providers vary, and Requesty points to each model's detail page for that provider's specific policy, so check the underlying provider policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-23: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety, and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and it points to each model's detail page for the underlying provider's retention and training policy. Routes can involve different model providers, so check the model detail page and that provider's own policy before sensitive use."
      }
    },
    {
      "id": "free-program-a2451835050e8aa84516",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-22: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante; Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-23: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is no longer in the free set and is listed at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-6e6e5d9b0f06d3a6c4f5",
      "program_id": "catalog2-merge-gateway",
      "event_type": "program_record_corrected",
      "program_name": "Merge Gateway free Nemotron route",
      "provider_label": "Merge Gateway",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://docs.merge.dev/merge-gateway/models/details/nvidia-nemotron-3-5-lightning-30b-a3b",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Merge's provider-owned catalog lists `nvidia/nemotron-3.5-lightning-30b-a3b` as available hosted text inference with one NVIDIA host and `Free / Free` input/output pricing. Merge's Free plan (pricing page, 2026-09-04) includes access to all free LLMs, automatic routing with built-in fallback, and starts without a credit card; the earlier $10 first-month credit was removed from the Free tier, while Pro adds $10 in expiring monthly credits and ordinarily charges model cost plus 5%. Merge also warns that its response `usage.cost` is provider cost rather than necessarily the final invoice, but the exact Nemotron route is explicitly cataloged as free rather than merely lacking pricing metadata."
      },
      "after": {
        "free_terms": "Merge's provider-owned catalog lists nvidia/nemotron-3.5-lightning-30b-a3b as available hosted text inference with one NVIDIA host and Free / Free input and output pricing, checked on 2026-09-23. Merge's Gateway pricing page puts access to all free LLMs, automatic routing with built-in fallback, and a start without a credit card in the Free plan, and lists all free models as included in every plan. The Free plan column shows no credit amount, while Merge's Get started documentation says the free plan comes with credit to spend on its model list, so treat any plan credit as an extra rather than as the thing paying for this route. Pro is a paid tier that charges LLM cost plus a 5% fee and issues $10 in complimentary credits each month that expire at the end of the month. Merge also warns that the usage.cost field in a response is provider cost rather than the invoiced amount, but this exact Nemotron route is cataloged as free rather than merely lacking pricing metadata."
      }
    },
    {
      "id": "free-program-6e095b8c211dafcdc3b9",
      "program_id": "catalog2-kenari",
      "event_type": "program_record_corrected",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-22) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen :free models are named in the Solo tier line on Kenari's homepage on 2026-09-22 (the same line read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog, and it changed again within a single day in September 2026).",
        "card": "No card is required for the Solo free lane: the plan card lists it as free with no top up, and the quickstart says no overseas credit card and no approval wait are needed. Optional top-up runs through QRIS from Rp 1.000 or crypto."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier that needs no top up and no card, with 50 requests per day across the current free-model roster (17 models on 2026-09-23) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Kenari's billing documentation states that the quota, per-minute rate, and top-up threshold in force at any moment are published on the pricing page and can be reset by the operator at any time, so confirm the current figures there before relying on them. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Seventeen provider-hosted :free text models per Kenari's homepage counter on 2026-09-23 (it also read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "No card is required. The Solo tier is listed as free with no top up, account creation asks only for an email address and password or a Google sign-in, and the quickstart says no overseas credit card is needed. Top-ups are optional and are paid by QRIS from Rp 1.000 or by crypto."
      }
    },
    {
      "id": "free-program-4ef0afe4a46038c395cc",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Models whose pricing tables mark standard Free tier input and output as free of charge, checked on 2026-09-18, are Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries: Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview no longer appear on the pricing page on 2026-09-18. This is model-specific: other listed models, including Gemini 3.1 Pro Preview, the Nano Banana image models, Gemini Omni Flash, Veo 3.1, Lyria, and Gemini 2.5 Computer Use Preview, and many discounted or batch modes say Free tier 'Not available.'"
      },
      "after": {
        "models_detail": "On 2026-09-23 the models whose pricing tables mark standard Free tier input and output as free of charge are Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries, Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription, both marked free of charge on 2026-09-04. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini Robotics ER 1.6 Preview, and a separate Gemini 2.5 Flash-Lite Preview entry no longer appear on the pricing page. This is model-specific: other listed models, including the Pro preview, image, video, and music models, show Free tier 'Not available', and discounted or batch modes may also say 'Not available' for models whose standard mode is free."
      }
    },
    {
      "id": "free-program-1759f1e3ed2edb75b5c7",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_terms_changed",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will notify users in advance by email if that status changes. Rate limits are applied per API key and cover requests as well as input and output tokens, and the API responds with HTTP Status Code 429 when any limit is exceeded. As of 2026-09-22 the Inference API documentation no longer publishes the numeric values for those limits, and Hetzner has said general inference rate limits were adjusted down during the experiment. Performance and availability are not guaranteed, requests can spend minutes waiting in queue at peak, and the platform should not be used for production environments."
      },
      "after": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will notify users in advance by email if that status changes. Rate limits are applied per API key and the API returns HTTP 429 when any of them is exceeded, but as of 2026-09-23 the Inference API docs page no longer publishes the numeric request and token figures. Performance and availability are not guaranteed, requests can queue at peak, and the platform is not for production use."
      }
    },
    {
      "id": "free-program-0493ecf0b1e00a63f97d",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-23T00:00:00+00:00",
      "date": "2026-09-23",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-22 the catalog listed 20 models, and OVHcloud describes the range on its product page as more than 20 models. The listed model series are Llama, Qwen, Qwen3Guard, GPT-OSS, Whisper, bge embeddings, NVIDIA Riva speech, and Stable Diffusion. Individual models can be retired under OVHcloud's lifecycle process, and on 2026-09-22 Mistral-7B-Instruct-v0.3 and Qwen3-Coder-30B-A3B-Instruct-FP8 were announced for decommissioning on 2026-10-01, with Mistral-Nemo-Instruct-2407 and Mistral-Small-3.2-24B-Instruct-2506 set for 2026-11-04. Seven catalog entries were priced at Free on 2026-09-22: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for English, German, Spanish and Italian. Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalogue. On 2026-09-23 the catalogue listed 20 models spanning embeddings, image generation, LLM guard, large language models, reasoning LLMs, speech to text, text to speech, and visual LLMs, drawn from the Llama, Qwen, GPT-OSS, Whisper, bge, NVIDIA Riva, and Stable Diffusion families. Individual models can be retired under OVHcloud's lifecycle process. Seven catalogue entries are priced at Free outright on 2026-09-23: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the nvr-tts-de-de, nvr-tts-es-es, nvr-tts-it-it, and nvr-tts-en-us text-to-speech voices. Qwen3.8-27B is paid at 0.4 EUR per million input tokens and 2.7 EUR per million output tokens."
      }
    },
    {
      "id": "free-program-da53d98382f35ffa2cf8",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page, which syncs prices in real time and shows available models only, usable within platform rate limits. The provider states that after verifying your identity you can use all free models, that free model calls are free and appear as a cost of 0 in the account bill, and that rate limits for free models are fixed values rather than rising with account spending. Model detail pages show each model's price and the highest speed limit available to users. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-20, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B (ChinaTelecom, listed at \uffe50 input and \uffe50 output, released on 2026-09-17), THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free on the same date: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 embedding and reranking models, the ChinaTelecom XingChen speech models, Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn, where the platform supports login via SMS or email, with the API key created in the console. The provider states that after verifying your identity you can use all free models."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page ('real-time prices, available models only'), usable within platform rate limits. Rate limits for the free models are fixed rather than tiered by spend, and per-model limits are listed on the model pages. Free model calls show as zero cost on the account bill. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-22, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, Qwen/Qwen2.5-7B-Instruct, and ChinaTelecom's XingChenAGI/Xing4.0-29B, which the model page lists at \u00a50 input and \u00a50 output. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, plus identity verification before the free models can be used, with the API key created in the console (SMS login favors Chinese numbers, email login supported)"
      }
    },
    {
      "id": "free-program-d932849df0541ed3d91c",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-21: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policy for the model serving a request comes from the underlying provider and is shown on each model detail page, so treatment varies by route: check Requesty and the underlying provider policies before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-22: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card is required on the Free plan.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policies of the underlying model providers vary, and Requesty points to each model's detail page for that provider's specific policy, so check the underlying provider policy before sensitive use."
      }
    },
    {
      "id": "free-program-9ae079f5b21484bd2db6",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_record_corrected",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "data_treatment"
      ],
      "before": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, counted per model; accounts that have topped up get twice the daily quota. A separate account-level rate limit also applies: accounts holding less than $5 in credit are capped at 20 requests per minute, while the 200 requests per minute shown in the docs' example rate-limit header is the limit for accounts holding $5 or more, not the free quota. The quota is counted in units that grow with prompt length, so long-context requests use it up faster. Past the quota, requests continue at the paid rate only when the account holds credit (a header can turn that fallback off); an account with no balance gets 429 until the quota resets.",
        "data_treatment": "Provider says it does not store message content by default; upstream provider terms still apply."
      },
      "after": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, counted per model; accounts that have topped up get twice the daily quota, and auto:free routing carries a separate parallel cap counted per IP address. The quota is counted in units that grow with prompt length, so long-context requests use it up faster. Past the quota, requests continue at the paid rate only when the account holds credit (a header can turn that fallback off); an account with no balance gets 429 until the quota resets. A separate account-wide rate limit also applies: 20 requests per minute while the account holds less than $5 of credit, and 200 requests per minute at $5 or more, with no daily cap at either level, so the example rate-limit header showing 200 requests per minute in the docs is the paid-tier figure.",
        "data_treatment": "Provider says it does not store message content by default: content is processed in memory and then discarded, while token counts, timestamps and model IDs are kept for billing. Every request passes automated content moderation, and the provider keeps the content of requests it judges to breach its acceptable use policy as compliance evidence. Messages are forwarded to upstream providers and remain subject to their privacy policies."
      }
    },
    {
      "id": "free-program-98b5bd52aec44923d23c",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-09-21) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Fifteen free models per the Solo plan card on Kenari's homepage on 2026-09-21, drawn from 80 models listed in total. Free models carry the :free suffix on the model id. Earlier counts were seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-22) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen :free models are named in the Solo tier line on Kenari's homepage on 2026-09-22 (the same line read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog, and it changed again within a single day in September 2026)."
      }
    },
    {
      "id": "free-program-7c4b4c818a91c3cf4904",
      "program_id": "kilo-code-free",
      "event_type": "program_terms_changed",
      "program_name": "Kilo :free model pool + kilo-auto/free router",
      "provider_label": "Kilo Code",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://kilo.ai/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Current provider-owned model documentation lists stepfun/step-3.7-flash:free, poolside/laguna-s-2.1:free, poolside/laguna-xs-2.1:free, nvidia/nemotron-3-ultra-550b-a55b:free, tencent/hy3:free, and openrouter/free. kilo-auto/free dynamically selects from a curated set whose membership changes as upstream availability changes. The free-account background route currently resolves to google/gemma-4-26b-a4b-it:free."
      },
      "after": {
        "models_detail": "Kilo's gateway documentation does not publish a fixed list of free model IDs and directs users to the live model catalog for current free options. Free models are identified by a :free suffix in the model ID, and on 2026-09-22 the authentication documentation gives minimax/minimax-m2.1:free and z-ai/glm-5:free as its examples. kilo-auto/free selects a model dynamically per session from a curated set of available free models, and that mapping updates server-side as free model availability shifts. For an account with no balance, the kilo-auto/small background route resolves to google/gemma-4-26b-a4b-it:free."
      }
    },
    {
      "id": "free-program-6ab1ff4aadd053a32e22",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Eight free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices seven models at Free for input and output tokens. Six of those also show Free cached-read pricing, and Jev 1.13 Free lists no cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-21: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of the Jev System One model from TypeSafe AI, called with the model id jev-1.13-free. Union Alpha Free left the free table by 2026-09-21, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, apart from the exceptions it lists. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the page warns against submitting personal or confidential data. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Jev 1.13 Free is not named among the exceptions.",
        "clock": "All seven are limited-time promotions; no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices eight models at Free for input and output tokens, and seven of those also show a free cached read. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-22: Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a structured decision model reached at the systemone endpoint rather than a chat model, and it is the one free row with no cached-read price. MiMo-V2.6-Flash Free joined the free table by 2026-09-22. Union Alpha Free left the free table by 2026-09-22, Muse Spark 1.2 Contributor Free left by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, with named exceptions. Big Pickle, MiMo-V2.6-Flash Free, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints whose use is logged for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Jev 1.13 Free is not named among the exceptions.",
        "clock": "All eight are limited-time promotions; no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-5ec1930438ca010fd6f5",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, per the program page on 2026-09-20. Solar Pro 2 is not named in the program.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved access starts at approval and lasts up to one year, and the program page caps the usage period at 2027-03-31, so an organization approved later in the window gets less than a full year."
      },
      "after": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, as named on the program page on 2026-09-22. Solar Pro 2 is not named in the program.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved organizations get access for up to one year, running from approval until 2027-03-31, so an organization approved later in the window gets less than a full year."
      }
    },
    {
      "id": "free-program-57f5a02eb149b2793a6b",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card and no paywall."
      },
      "after": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority."
      }
    },
    {
      "id": "free-program-56e8064d5d8269a3d82b",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-21 the public catalog lists 20 models and OVHcloud describes the range as more than 20 models, covering Llama, Qwen, GPT-OSS, Whisper speech to text, NVIDIA Riva text to speech, bge and Qwen embedding, and Stable Diffusion image generation entries. Several catalog entries are priced at Free outright, on 2026-09-21 the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four nvr-tts voices, while Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens. Individual models can be retired under OVHcloud's lifecycle process: four models announced for decommissioning on 2026-07-02 and 2026-08-04 stay reachable through the API during their grace period but are hidden from the website catalog, so the listed count understates what is still callable."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-22 the catalog listed 20 models, and OVHcloud describes the range on its product page as more than 20 models. The listed model series are Llama, Qwen, Qwen3Guard, GPT-OSS, Whisper, bge embeddings, NVIDIA Riva speech, and Stable Diffusion. Individual models can be retired under OVHcloud's lifecycle process, and on 2026-09-22 Mistral-7B-Instruct-v0.3 and Qwen3-Coder-30B-A3B-Instruct-FP8 were announced for decommissioning on 2026-10-01, with Mistral-Nemo-Instruct-2407 and Mistral-Small-3.2-24B-Instruct-2506 set for 2026-11-04. Seven catalog entries were priced at Free on 2026-09-22: the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four NVIDIA Riva text-to-speech voices for English, German, Spanish and Italian. Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens."
      }
    },
    {
      "id": "free-program-55e9f555d76c1cc1d7eb",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-21: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-22: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante; Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is priced at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-4ddc7cb58c1a5bacaf49",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_terms_changed",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will email in advance if that changes. Published limits per API key: 10 requests per 60 seconds, 4M input tokens and 100K output tokens per 60 seconds, HTTP 429 past any of them. Performance and availability are not guaranteed, requests can queue at peak, and the platform is not for production use."
      },
      "after": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will notify users in advance by email if that status changes. Rate limits are applied per API key and cover requests as well as input and output tokens, and the API responds with HTTP Status Code 429 when any limit is exceeded. As of 2026-09-22 the Inference API documentation no longer publishes the numeric values for those limits, and Hetzner has said general inference rate limits were adjusted down during the experiment. Performance and availability are not guaranteed, requests can spend minutes waiting in queue at peak, and the platform should not be used for production environments."
      }
    },
    {
      "id": "free-program-49ee998718cadc6f4490",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-22T00:00:00+00:00",
      "date": "2026-09-22",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single pool of 1M tokens per day shared across all free models, reset every day. Some models carry lower request caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute for the Gemini image preview model, and some models count one call as several requests. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "58 models listed at $0 input and $0 output on the provider's free catalog page (read 2026-09-17), nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "card": "No credit card is required for the 10 trial calls, which never expire. The recurring daily quotas are reachable only after a one-time top-up of at least $1, so a payment is needed once; free-model usage then bills $0 for input and output and does not draw that balance down."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls, shared across all free models, with no credit card and no expiry. A one-time top-up of at least $1 switches the account permanently to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across the catalog and reset every day. Individual models can carry lower request caps, stepping down through 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute, because a call to a heavier model counts as several requests (caps read on 2026-09-22). Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "60 models listed at $0 input and $0 output on the provider's free catalog page, which the provider marked updated on 2026-09-21: nearly all text models, plus two that return images (gpt-image-2-free and gemini-3.1-flash-image-preview-free). The roster changes over time.",
        "card": "No credit card is required to open an account or to use the 10 trial calls. Moving to the permanent daily quotas requires a one-time top-up of at least $1, which is a payment rather than a card kept on file."
      }
    },
    {
      "id": "free-program-e995c2faa59afd78b2a3",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority."
      },
      "after": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card and no paywall."
      }
    },
    {
      "id": "free-program-e26dd99a5f830d985527",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-20: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan states that no credit card is required.",
        "data_treatment": "Requesty says it never uses customer data for training, and points to each model's detail page for that provider's retention and training policy, so treatment varies with the underlying model provider. Check Requesty and the underlying provider policies before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-21: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training. Retention and training policy for the model serving a request comes from the underlying provider and is shown on each model detail page, so treatment varies by route: check Requesty and the underlying provider policies before sensitive use."
      }
    },
    {
      "id": "free-program-a90196eaed9df50c44a6",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-09-20) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Fifteen free models on Kenari's homepage Solo card on 2026-09-20, identified by the :free suffix and billed Rp 0 (the count read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "Kenari's homepage lists the Solo tier as free with no top up, so no payment card is needed for the free lane. Top-ups, when a user chooses one, go through QRIS from Rp 1,000 and require no foreign credit card."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-09-21) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Fifteen free models per the Solo plan card on Kenari's homepage on 2026-09-21, drawn from 80 models listed in total. Free models carry the :free suffix on the model id. Earlier counts were seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27, so the roster moves with the provider catalog.",
        "card": "No card is required for the Solo free lane: the plan card lists it as free with no top up, and the quickstart says no overseas credit card and no approval wait are needed. Optional top-up runs through QRIS from Rp 1.000 or crypto."
      }
    },
    {
      "id": "free-program-a05db9e719e4a7fe3ae3",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Seven free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices seven models at Free. Six of them are Free for input, output, and cached-read tokens, and Jev 1.13 Free is Free for input and output with no cached-read price listed. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-20: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a System One model from TypeSafe AI, called with the model id jev-1.13-free on Zen's systemone endpoint; the paid Jev 1.13 row is priced at $0.042 per 1M input tokens. Union Alpha Free is absent from both the pricing table and the endpoints table on 2026-09-20. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints that log use for security and to improve NVIDIA products and services, and the provider says not to submit personal or confidential data to them. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Jev 1.13 Free is not named among the stated exceptions, so the general policy applies to it: providers follow a zero-retention policy and do not use your data for model training."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices seven models at Free for input and output tokens. Six of those also show Free cached-read pricing, and Jev 1.13 Free lists no cached-read price. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-21: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is the free version of the Jev System One model from TypeSafe AI, called with the model id jev-1.13-free. Union Alpha Free left the free table by 2026-09-21, Muse Spark 1.2 Contributor Free left it by 2026-09-17, and Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Zen states that its providers follow a zero-retention policy and do not use customer data for model training, apart from the exceptions it lists. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security and to improve NVIDIA products and services, and the page warns against submitting personal or confidential data. Muse Spark 1.3 Contributor Free trades discounted token pricing for permission to use prompts and completions to train future Meta models. Jev 1.13 Free is not named among the exceptions."
      }
    },
    {
      "id": "free-program-8197416960a8c26615f2",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Mancer's current live model table assigns both input and output token cost as FREE for MythoLite. Other currently listed models have positive per-token credit prices.",
        "models_detail": "MythoLite only: a LLaMA 2 13B creative/roleplay model priced FREE for input and output on the current model table. Its context and output caps sit behind a details fold the nightly read does not expand, so they are not stated here."
      },
      "after": {
        "free_terms": "On 2026-09-21 Mancer's model table lists MythoLite as the only model priced FREE for both input and output tokens. Mancer's pricing FAQ says that some models are free while most have a per token credit cost, and it presents the free models as a way to test the API before making a purchase. Free use does not depend on a paid balance: Mancer states that a user can still use free models with a negative credit count. MythoMax, the other LLaMA 2 13B model on the list, is priced at 0.14 credits per input token and 0.24 credits per output token.",
        "models_detail": "MythoLite only: on 2026-09-21 the model table lists it as a LLaMA 2 13B model tagged Creative, Roleplay, with input and output token cost both shown as FREE. The listing does not state MythoLite's context window or maximum output length."
      }
    },
    {
      "id": "free-program-69214cd44528912e984c",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries stood at $0.00/1M on 2026-09-20: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M."
      },
      "after": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-21: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-621619ccee20e031eafa",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model. OVHcloud's product page describes over 40 models, such as Llama, Qwen and DeepSeek, but the public catalog listed 20 models on 2026-09-17: Qwen3.8-27B, Qwen3.6-27B, Qwen3.5-397B-A17B, Qwen3.5-9B, Qwen3-Embedding-8B, Qwen3Guard-Gen-8B, Qwen3Guard-Gen-0.6B, gpt-oss-120b, gpt-oss-20b, whisper-large-v3, whisper-large-v3-turbo, Meta-Llama-3_3-70B-Instruct, Qwen2.5-VL-72B-Instruct, bge-multilingual-gemma2, bge-m3, stable-diffusion-xl-base-v10, and four NVIDIA Riva text-to-speech voices. Models announced for retirement stay callable through the API but are hidden from the catalog. The authenticated catalog also prices a few entries at Free outright, on 2026-09-17 the beta Qwen3Guard-Gen 0.6B and 8B moderation models (Qwen3.8-27B itself is paid at 0.4 EUR in / 2.7 EUR out per 1M tokens)."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. On 2026-09-21 the public catalog lists 20 models and OVHcloud describes the range as more than 20 models, covering Llama, Qwen, GPT-OSS, Whisper speech to text, NVIDIA Riva text to speech, bge and Qwen embedding, and Stable Diffusion image generation entries. Several catalog entries are priced at Free outright, on 2026-09-21 the beta Qwen3Guard-Gen 0.6B and 8B moderation models, stable-diffusion-xl-base-v10, and the four nvr-tts voices, while Qwen3.8-27B is paid at 0.4 EUR in and 2.7 EUR out per 1M tokens. Individual models can be retired under OVHcloud's lifecycle process: four models announced for decommissioning on 2026-07-02 and 2026-08-04 stay reachable through the API during their grace period but are hidden from the website catalog, so the listed count understates what is still callable."
      }
    },
    {
      "id": "free-program-2379e2058658db1f9304",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_record_corrected",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 then switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, reset every day. Most free models carry the 100 per day and 10 per minute cap, but models with a higher request weight are capped lower, as listed on 2026-09-20: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute for gpt-4.1-free and gpt-4o-free, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the daily free quotas requires a one-time top-up of at least $1, which free usage does not draw down, and the reviewed provider pages do not state which payment methods that top-up accepts.",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and a single pool of 1M tokens per day shared across all free models, reset every day. Some models carry lower request caps, ranging from 50 per day and 5 per minute down to 10 per day and 1 per minute for the Gemini image preview model, and some models count one call as several requests. Past a cap the API returns 429 until the next minute or day starts.",
        "card": "No credit card is required for the 10 trial calls, which never expire. The recurring daily quotas are reachable only after a one-time top-up of at least $1, so a payment is needed once; free-model usage then bills $0 for input and output and does not draw that balance down.",
        "card_required": "yes"
      }
    },
    {
      "id": "free-program-1fe772493870c5bc3685",
      "program_id": "ibm-watsonx-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "watsonx.ai foundation models (playground)",
      "provider_label": "IBM",
      "detected_at": "2026-09-21T00:00:00+00:00",
      "date": "2026-09-21",
      "evidence_url": "https://www.ibm.com/products/watsonx-ai/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "clock"
      ],
      "before": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan with 300,000 foundation-model inference tokens per month, 20 CUH per month for supported machine-learning workloads, 100 text-extraction pages per month, and a plan-level limit of 2 inference requests per second. Foundation-model tuning and custom foundation models are unavailable on Lite.",
        "models_detail": "Ready-to-use IBM and third-party foundation models available to the Lite instance, subject to regional and current catalog availability. IBM's supported-model surface describes IBM and third-party models; custom and deploy-on-demand foundation models are not included in Lite. An exact immutable Lite model roster is not stated.",
        "clock": "Monthly allowance; no fixed offer end date stated."
      },
      "after": {
        "free_terms": "Standing watsonx.ai Runtime Lite plan, a free quota based service plan with 300,000 foundation model inference tokens per month, 20 capacity unit hours per month for machine learning workloads, and 100 document text classification and extraction pages per month, with a plan level limit of 2 inference requests per second. Custom foundation models are not available on the Lite plan. Reaching any Lite plan quota suspends the service for the remainder of that month.",
        "models_detail": "Ready to use IBM and third party foundation models are available to the Lite instance, subject to regional and current catalog availability. IBM's supported models page describes both IBM and third party foundation models provided with watsonx.ai for inferencing, and states that foundation model availability varies by data center location. Custom foundation models are not available on the Lite plan. An exact immutable Lite model roster is not stated.",
        "clock": "Monthly allowance with quota limits that reset on the first of every month. No fixed offer end date is stated."
      }
    },
    {
      "id": "free-event-29af961f1ee6c4d0279b",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "deepseek/deepseek-v4-flash-0731:free",
      "model": "DeepSeek: DeepSeek V4 Flash 0731",
      "model_key": "deepseekv4flash0731free",
      "detected_at": "2026-09-20T11:30:17+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://openrouter.ai/deepseek/deepseek-v4-flash-0731:free"
    },
    {
      "id": "free-program-f2f352cd1e9e645fd1af",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, counted per model; accounts that have topped up get twice the daily quota. The quota is counted in units that grow with prompt length, so long-context requests use it up faster. Past the quota, requests continue at the paid rate only when the account holds credit (a header can turn that fallback off); an account with no balance gets 429 until the quota resets. The docs' example rate-limit header (200 requests per minute) is the paid-tier account limit, not the free quota."
      },
      "after": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, counted per model; accounts that have topped up get twice the daily quota. A separate account-level rate limit also applies: accounts holding less than $5 in credit are capped at 20 requests per minute, while the 200 requests per minute shown in the docs' example rate-limit header is the limit for accounts holding $5 or more, not the free quota. The quota is counted in units that grow with prompt length, so long-context requests use it up faster. Past the quota, requests continue at the paid rate only when the account holds credit (a header can turn that fallback off); an account with no balance gets 429 until the quota resets."
      }
    },
    {
      "id": "free-program-ed862116463da3461ea3",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication method bills usage to an AWS account. No numeric limits are published publicly.",
        "card": "Not stated by the provider."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). The Amazon Nova Act user guide, read on 2026-09-20, states that API key authentication using a key from nova.amazon.com/act tied to an Amazon.com account is free of charge with daily limits, while a separate AWS IAM authentication route carries usage charges on an AWS account. No numeric limits are published publicly.",
        "card": "Not required. The Amazon Nova Act user guide states that API key authentication is free of charge with daily limits, and no payment terms appear anywhere in the Nova Terms of Use. Charges apply only on the separate AWS IAM route, which is not this free offering."
      }
    },
    {
      "id": "free-program-ea76d7c3f0c8880659c3",
      "program_id": "groq-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Groq free-tier models",
      "provider_label": "Groq",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://console.groq.com/docs/rate-limits.md",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-09-02: groq/compound and groq/compound-mini 30 RPM, 250 RPD, 70K TPM; openai/gpt-oss-120b, gpt-oss-20b, gpt-oss-safeguard-20b, qwen/qwen3.6-27b, and qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and -86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio-seconds per hour, 28.8K audio-seconds per day. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table as of 2026-09-01: groq/compound, groq/compound-mini, openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.6-27b, qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      },
      "after": {
        "free_terms": "Standing organization-level Free Plan with no usage charges and model-specific recurring rate limits. Free Plan table on 2026-09-20: openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, and qwen/qwen3.8-27b 30 RPM, 1K RPD, 8K TPM, 200K TPD; meta-llama/llama-prompt-guard-2-22m and -86m 30 RPM, 14.4K RPD, 15K TPM, 500K TPD; canopylabs/orpheus-arabic-saudi and orpheus-v1-english 10 RPM, 100 RPD, 1.2K TPM, 3.6K TPD; whisper-large-v3 and whisper-large-v3-turbo 20 RPM, 2K RPD, 7.2K audio-seconds per hour, 28.8K audio-seconds per day. The page says the exact current limits are on the authenticated account Limits page; a valid payment method is needed only to upgrade to the Developer plan.",
        "models_detail": "Free Plan rate-limit table as of 2026-09-20: canopylabs/orpheus-arabic-saudi, canopylabs/orpheus-v1-english, meta-llama/llama-prompt-guard-2-22m, meta-llama/llama-prompt-guard-2-86m, openai/gpt-oss-120b, openai/gpt-oss-20b, openai/gpt-oss-safeguard-20b, qwen/qwen3.8-27b, whisper-large-v3, and whisper-large-v3-turbo. Removed from the table since 2026-09-02: groq/compound, groq/compound-mini, and qwen/qwen3.6-27b. Removed earlier, since 2026-08-25: llama-3.1-8b-instant, llama-3.3-70b-versatile, llama-4-scout, and llama-guard-4-12b."
      }
    },
    {
      "id": "free-program-e59f105daea82ff5110c",
      "program_id": "catalog2-regolo-ai",
      "event_type": "program_record_corrected",
      "program_name": "Regolo brick-v1-beta at zero price",
      "provider_label": "Regolo AI",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://regolo.ai/pricing/",
      "changed_fields": [
        "data_treatment"
      ],
      "before": {
        "data_treatment": "Not stated on the reviewed pricing, model, and catalog pages; check provider privacy terms before sensitive use."
      },
      "after": {
        "data_treatment": "Regolo's pricing page presents the service under a Zero Data Retention banner and describes its plans as zero-retention AI. The reviewed pricing, model, and catalog pages do not say whether prompts or outputs are used for training, so check the provider privacy terms before sensitive use."
      }
    },
    {
      "id": "free-program-e4d65b2e72a0ab63d8f7",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_record_corrected",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "data_treatment"
      ],
      "before": {
        "data_treatment": "Hetzner's blog states 'We do not log any prompt data.'"
      },
      "after": {
        "data_treatment": "Hetzner's Inference API FAQ says it keeps only the data needed to track usage and, in the future, bill usage, naming request timestamps and token counts. The same FAQ says, 'We do not store the content of request and response and we do not plan on doing so in the future (unless some law requires us to do so).' Hetzner's engineering blog states, 'We do not log any prompt data.'"
      }
    },
    {
      "id": "free-program-e35be6d1d8b13f44d2cd",
      "program_id": "ai-horde-community-free",
      "event_type": "program_record_corrected",
      "program_name": "AI Horde community inference (anonymous)",
      "provider_label": "AI Horde",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://aihorde.net/",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "not_stated"
      },
      "after": {
        "card": "No credit card is required. AI Horde states that everyone can use the service for free, even anonymously, with no credit card, no paywall, and no way to buy priority."
      }
    },
    {
      "id": "free-program-cc4982b0c2c9c465ad15",
      "program_id": "catalog2-requesty",
      "event_type": "program_record_corrected",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-18: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and that each model detail page carries the data retention and training policy of the provider serving that model, so treatment varies by route. Check both the Requesty policy and the underlying provider policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-20: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The docs page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan states that no credit card is required.",
        "data_treatment": "Requesty says it never uses customer data for training, and points to each model's detail page for that provider's retention and training policy, so treatment varies with the underlying model provider. Check Requesty and the underlying provider policies before sensitive use."
      }
    },
    {
      "id": "free-program-c976068aacda752407ea",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup"
      ],
      "before": {
        "free_terms": "Standing free-model roster on the CN pricing page ('real-time prices, available models only'), usable within platform rate limits. Per-model rate limits are shown on each model detail page after login. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-18, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B, THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, the ChinaTelecom XingChen speech models, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn, where the platform supports login via SMS or email, with the API key created in the console (SMS login favors Chinese numbers, email login supported). The provider states that after verifying your identity you can use all free models, so identity verification is part of reaching the free roster."
      },
      "after": {
        "free_terms": "Standing free-model roster on the CN pricing page, which syncs prices in real time and shows available models only, usable within platform rate limits. The provider states that after verifying your identity you can use all free models, that free model calls are free and appear as a cost of 0 in the account bill, and that rate limits for free models are fixed values rather than rising with account spending. Model detail pages show each model's price and the highest speed limit available to users. The roster is dynamically maintained by the provider and models can be added or removed; the provider states it retains interpretation rights over which models are free.",
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-20, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B (ChinaTelecom, listed at \uffe50 input and \uffe50 output, released on 2026-09-17), THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free on the same date: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), the BAAI bge-m3, bge-reranker-v2-m3, bge-large-zh-v1.5 and bge-large-en-v1.5 embedding and reranking models, the ChinaTelecom XingChen speech models, Qwen/Qwen3-ASR-1.7B, FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn, where the platform supports login via SMS or email, with the API key created in the console. The provider states that after verifying your identity you can use all free models."
      }
    },
    {
      "id": "free-program-abfd448f1fc069f43f29",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of $1 or more permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across all free models and reset every day. Some models carry a higher request weight, so one call counts as several requests and their caps are lower: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free (caps read on 2026-09-17). Past a cap the API returns 429 until the next minute or day starts. Free usage bills $0 and does not draw the topped-up balance down.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the recurring daily quotas requires a one-time top-up of $1 or more, which is a single payment rather than a card kept on file, and the provider does not state which payment methods it accepts. Free usage bills $0 and does not draw that balance down.",
        "card_required": "yes"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 then switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and one shared pool of 1M tokens per day, reset every day. Most free models carry the 100 per day and 10 per minute cap, but models with a higher request weight are capped lower, as listed on 2026-09-20: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute for gpt-4.1-free and gpt-4o-free, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free. Past a cap the API returns 429 until the next minute or day starts.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the daily free quotas requires a one-time top-up of at least $1, which free usage does not draw down, and the reviewed provider pages do not state which payment methods that top-up accepts.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-7a769379cc3f44c2173d",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Seven free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices seven models at Free for input, output, and cached-read tokens. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the current Zen pricing table: Big Pickle, Union Alpha Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, and Muse Spark 1.3 Contributor Free. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Union Alpha Free's provider states a zero-retention policy; the Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and the provider says not to submit personal or confidential data. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Union Alpha Free's provider follows a zero-retention policy and does not use your data for model training."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices seven models at Free. Six of them are Free for input, output, and cached-read tokens, and Jev 1.13 Free is Free for input and output with no cached-read price listed. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-20: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free, and Jev 1.13 Free. Jev 1.13 Free is a System One model from TypeSafe AI, called with the model id jev-1.13-free on Zen's systemone endpoint; the paid Jev 1.13 row is priced at $0.042 per 1M input tokens. Union Alpha Free is absent from both the pricing table and the endpoints table on 2026-09-20. Muse Spark 1.2 Contributor Free left the free table by 2026-09-17; Ox Alpha Free and Hy3 Free were removed before 2026-09-01. The Nemotron endpoints are NVIDIA trial endpoints whose use is logged. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are trial endpoints that log use for security and to improve NVIDIA products and services, and the provider says not to submit personal or confidential data to them. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Jev 1.13 Free is not named among the stated exceptions, so the general policy applies to it: providers follow a zero-retention policy and do not use your data for model training."
      }
    },
    {
      "id": "free-program-517b8dec0ce8e6db9fec",
      "program_id": "catalog2-zenmux",
      "event_type": "program_record_corrected",
      "program_name": "ZenMux GLM Flash free routes",
      "provider_label": "ZenMux",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://zenmux.ai/z-ai/glm-4.6v-flash-free",
      "changed_fields": [
        "free_terms",
        "signup"
      ],
      "before": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as Active routes, each labeled Free and Rate Limit, with $0 input and $0 output pricing shown on the provider rows. Access requires a ZenMux account, which can be created with email, a GitHub account, or a Google account, plus an API key created after signing in. ZenMux offers two usage plans, Pay As You Go and the Builder Plan subscription starting at $20 per month, and the reviewed pages do not state that either plan is required to call these free routes. The pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, and that some models require a positive balance, without stating that these two free routes are among them.",
        "signup": "A ZenMux account is required and can be created with email, a GitHub account, or a Google account, after which an API key is created from the management page. Account balance must remain non-negative. The reviewed pages do not state that a card, a funded balance, or a particular usage plan is required for these active free routes."
      },
      "after": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active $0/$0 routes on 2026-09-20, each marked Free and Rate Limit. Access requires a ZenMux account and an API key created after signing in. ZenMux offers a Pay As You Go plan and a paid Builder subscription, and the public pages do not state which key type reaches these free routes, nor do they state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance. Prior free routes have sunset without becoming permanent, so the lineup is dynamic.",
        "signup": "A ZenMux account and an API key are required, and ZenMux accepts email, GitHub, or Google sign in. The account balance must remain non-negative for all models including free tiers, and some models require a positive balance. The public pages do not state a card requirement or a positive funded balance for these active free routes."
      }
    },
    {
      "id": "free-program-42815e7c40496478181d",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "models_detail",
        "clock"
      ],
      "before": {
        "models_detail": "Solar Pro 2 and Solar Pro 3 (LLM) plus Document Parse, per the program page. The current flagship Solar Pro 4 is not named in the program.",
        "clock": "The program page lists the event period as running through March 31, 2027 11:00 PM (KST). Approved organizations have free access from the date of approval until March 31, 2027, for up to one year, so that date caps access rather than only the application window."
      },
      "after": {
        "models_detail": "Solar Pro 3 and Solar Pro 4 (LLM) plus Document Parse, per the program page on 2026-09-20. Solar Pro 2 is not named in the program.",
        "clock": "Applications are accepted through 2027-03-31 23:00 KST. Approved access starts at approval and lasts up to one year, and the program page caps the usage period at 2027-03-31, so an organization approved later in the window gets less than a full year."
      }
    },
    {
      "id": "free-program-11ea49e2d1d62f587763",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-18: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.11 in and $0.44 out per 1M."
      },
      "after": {
        "models_detail": "Seven catalog entries stood at $0.00/1M on 2026-09-20: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.13 in and $0.53 out per 1M."
      }
    },
    {
      "id": "free-program-0a4129b825c57d7ccd48",
      "program_id": "routelm-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "RouteLM monthly free tokens",
      "provider_label": "RouteLM",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://routelm.dev/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "The 50,000 monthly free tokens apply only to the eight models the pricing table marks Eligible: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. The rest of the catalog (Claude, GPT, Grok, Kimi, MiniMax, GLM-5.2, DeepSeek V4 Pro) is marked Standard and is paid only.",
        "signup": "required; email + password account with email verification and Turnstile CAPTCHA human verification"
      },
      "after": {
        "models_detail": "The 50,000 monthly free tokens apply only to the models the pricing table marks Eligible. On 2026-09-20 that list held eight models: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. Every named model outside that list (Claude, GPT, Grok, Kimi, MiniMax, GLM-5.2, DeepSeek V4 Pro) is marked Standard and is paid only. The auto smart router is listed separately with dynamic input and output costs and a free tier quota that varies by the model it routes to.",
        "signup": "required; account created with full name, email address, and password, followed by email verification and Turnstile CAPTCHA human verification"
      }
    },
    {
      "id": "free-program-05c10c0b365701c76fc6",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-20T00:00:00+00:00",
      "date": "2026-09-20",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-18) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models per the Solo plan card on Kenari's homepage on 2026-09-18 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog). Kenari's billing documentation says the operator can change the quota, rate limit, and top-up threshold at any time, so the current values live on the pricing page.",
        "card": "not_stated"
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (15 models on 2026-09-20) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Fifteen free models on Kenari's homepage Solo card on 2026-09-20, identified by the :free suffix and billed Rp 0 (the count read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "card": "Kenari's homepage lists the Solo tier as free with no top up, so no payment card is needed for the free lane. Top-ups, when a user chooses one, go through QRIS from Rp 1,000 and require no foreign credit card."
      }
    },
    {
      "id": "free-event-1f733b7abf74e8447f96",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "deepseek/deepseek-v4-flash-0731:free",
      "model": "DeepSeek: DeepSeek V4 Flash 0731",
      "model_key": "deepseekv4flash0731free",
      "detected_at": "2026-09-18T19:01:18+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://openrouter.ai/deepseek/deepseek-v4-flash-0731:free"
    },
    {
      "id": "free-event-0c73e132df15ab6737a3",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "qwen/qwen3.8-27b:free",
      "model": "Qwen: Qwen3.8 27B",
      "model_key": "qwen3827bfree",
      "detected_at": "2026-09-18T19:01:18+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://openrouter.ai/qwen/qwen3.8-27b:free"
    },
    {
      "id": "free-program-eab519cb268b3b35501d",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "The Amazon Nova Act developer guide says API key access from nova.amazon.com/act is free of charge with daily limits. The daily limit numbers are not published on these pages. Under the Nova.amazon.com Terms of Use (last updated December 2, 2025), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and some software requires an Amazon-provided API key (Terms 1.5). The Build with Nova schedule describes an environment for experimentation that does not support deploying production-scale developer offerings (S-1.1). For production use, AWS IAM authentication is recommended, and usage charges apply to the AWS account."
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). Amazon's Nova Act user guide states that API key authentication is free of charge with daily limits, while the separate AWS IAM authentication method bills usage to an AWS account. No numeric limits are published publicly."
      }
    },
    {
      "id": "free-program-aa6d42101f86c515d834",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Seven free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "data_treatment"
      ],
      "before": {
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free and Muse Spark 1.2 Contributor Free permit prompts and completions to train future Meta models."
      },
      "after": {
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free are NVIDIA trial endpoints: use is logged for security purposes and to improve NVIDIA products and services, and the provider says not to submit personal or confidential data. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Union Alpha Free's provider follows a zero-retention policy and does not use your data for model training."
      }
    },
    {
      "id": "free-program-a67ca97d8e29d42b4244",
      "program_id": "catalog2-zenmux",
      "event_type": "program_record_corrected",
      "program_name": "ZenMux GLM Flash free routes",
      "provider_label": "ZenMux",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://zenmux.ai/z-ai/glm-4.6v-flash-free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as active $0/$0 routes, each marked Free and Rate Limit. Access requires a ZenMux account and PAYG API key; the public pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, while some models can require a positive balance. Prior free routes have sunset without becoming permanent, so the lineup is dynamic.",
        "models_detail": "Currently active free routes: z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free. Earlier Kimi free routes are marked Sunset and are not included.",
        "signup": "ZenMux account and PAYG API key required. Account balance must remain non-negative; reviewed pages do not prove a card or positive funded balance is required for these active free routes.",
        "clock": "No fixed expiry for the two active routes; lineup is dynamic and prior free routes have sunset without becoming permanent."
      },
      "after": {
        "free_terms": "ZenMux lists z-ai/glm-4.6v-flash-free and z-ai/glm-4.7-flash-free as Active routes, each labeled Free and Rate Limit, with $0 input and $0 output pricing shown on the provider rows. Access requires a ZenMux account, which can be created with email, a GitHub account, or a Google account, plus an API key created after signing in. ZenMux offers two usage plans, Pay As You Go and the Builder Plan subscription starting at $20 per month, and the reviewed pages do not state that either plan is required to call these free routes. The pages do not state a card requirement or a numeric free-route quota. ZenMux documents that an account must keep a non-negative balance even for free tiers, and that some models require a positive balance, without stating that these two free routes are among them.",
        "models_detail": "On 2026-09-18 the active free routes are z-ai/glm-4.6v-flash-free, published 2025-12-08, and z-ai/glm-4.7-flash-free, published 2026-01-19. ZenMux changes its free lineup over time, so confirm a route on its own model page before relying on it.",
        "signup": "A ZenMux account is required and can be created with email, a GitHub account, or a Google account, after which an API key is created from the management page. Account balance must remain non-negative. The reviewed pages do not state that a card, a funded balance, or a particular usage plan is required for these active free routes.",
        "clock": "The two active routes carry no stated expiry date. ZenMux changes its free lineup over time, so a route can be withdrawn without notice."
      }
    },
    {
      "id": "free-program-9bcb3eac5743f688ce77",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (17 models on 2026-09-17) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Seventeen provider-hosted :free text models per Kenari's homepage counter on 2026-09-17 (it read eleven on 2026-09-04 and twelve on 2026-08-27; the roster moves with the provider catalog)."
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (16 models on 2026-09-18) at up to 5 requests per minute. A cumulative top-up of Rp 10,000 raises that to 1,000 requests per day. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee; the daily quota returns HTTP 429 with a Retry-After header when exhausted.",
        "models_detail": "Sixteen free models per the Solo plan card on Kenari's homepage on 2026-09-18 (it read seventeen on 2026-09-17, eleven on 2026-09-04, and twelve on 2026-08-27; the roster moves with the provider catalog). Kenari's billing documentation says the operator can change the quota, rate limit, and top-up threshold at any time, so the current values live on the pricing page."
      }
    },
    {
      "id": "free-program-777f35c8f3f6f9b5f942",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_record_corrected",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "card",
        "clock",
        "card_required",
        "access_clock"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, resetting daily (a few models are capped at 50 per day and 5 per minute).",
        "card": "not required for the 10 trial calls; a one-time $1 top-up (not a card on file) unlocks the daily free quotas",
        "clock": "Ten no-expiry trial calls per new account; daily free quotas after a one-time $1 top-up, with no published end date.",
        "card_required": "no",
        "access_clock": "limited_undated"
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of $1 or more permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day, shared across all free models and reset every day. Some models carry a higher request weight, so one call counts as several requests and their caps are lower: 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for gemini-3.1-flash-image-preview-free (caps read on 2026-09-17). Past a cap the API returns 429 until the next minute or day starts. Free usage bills $0 and does not draw the topped-up balance down.",
        "card": "No credit card is required for the 10 trial calls, and those calls never expire. Reaching the recurring daily quotas requires a one-time top-up of $1 or more, which is a single payment rather than a card kept on file, and the provider does not state which payment methods it accepts. Free usage bills $0 and does not draw that balance down.",
        "clock": "The 10 trial calls never expire. After a one-time top-up of $1 or more, the daily quotas are permanent and reset every day, and the provider publishes no end date.",
        "card_required": "yes",
        "access_clock": "standing"
      }
    },
    {
      "id": "free-program-5725e627f9e4beea7db4",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-17: NVIDIA Nemotron 3.5 Lightning 30B A3B, Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, and Nemotron 3.5 Content Safety; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny; and Meta Muse Glimmer 30B. The docs page carries a shorter nine-row table. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card required for the Free plan.",
        "data_treatment": "Requesty says it never uses your data for training. Data retention and training policy for each model depends on the underlying provider and is shown on that model's detail page, so check it before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-18: NVIDIA Nemotron 3 Ultra 550B A55B, Nemotron 3 Super 120B A12B, Nemotron 3 Nano 30B A3B, Nemotron 3 Nano Omni 30B A3B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B A3B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B IT; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that leaves out Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "The Free plan requires no credit card.",
        "data_treatment": "Requesty says it never uses customer data for training, and that each model detail page carries the data retention and training policy of the provider serving that model, so treatment varies by route. Check both the Requesty policy and the underlying provider policy before sensitive use."
      }
    },
    {
      "id": "free-program-4a983d81a3935a985909",
      "program_id": "catalog2-unorouter",
      "event_type": "program_record_corrected",
      "program_name": "UnoRouter free models",
      "provider_label": "UnoRouter",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://unorouter.com/en/docs/platform/models-and-pricing",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock",
        "card_required"
      ],
      "before": {
        "free_terms": "UnoRouter offers $0-input and $0-output models. Enrollment requires a UnoRouter account through Discord or GitHub plus an API key; no credit card is required. UnoRouter's own quota is approximately one request per minute per free model per user, layered over upstream providers' per-minute, daily-token, and shared-pool limits. Exhausted routes return 429, or 503 when every provider is busy. There is no announced expiry, but availability is best-effort and the model set shifts over time.",
        "models_detail": "Models currently marked free in UnoRouter's own catalog; the set changes as shared pools drain or upstream providers change.",
        "signup": "UnoRouter account through Discord or GitHub plus API key required; no credit card required.",
        "clock": "Standing best-effort catalog with approximately 1 request/minute per free model plus upstream shared-pool limits; no published expiry.",
        "card_required": "no"
      },
      "after": {
        "free_terms": "UnoRouter offers models priced at $0 for input and $0 for output, marked with a :free suffix. A :free model routes only to free providers and never uses an account balance, and the same name without the suffix is the paid version. Using the API requires a UnoRouter account and an API key. UnoRouter's own cap is 1 request per minute, per free model, per user, layered over the upstream provider's own cap, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps triggered by very large prompts, and a per-user concurrency limit on parallel requests. Exhausted routes return 429, and 503 when every provider for a model is busy. There is no announced expiry, but availability is best effort and the model set shifts over time.",
        "models_detail": "Free models carry a :free suffix, for example gpt-oss-120b:free, and the paid version is the same model name without the suffix. Free models were listed in UnoRouter's own catalog on 2026-09-18, and the set changes as shared pools drain or upstream providers change. Not all free models support tool calls or vision, so check the capability badges on the model page.",
        "signup": "A UnoRouter account and an API key are required. The sign-in methods offered and any payment method requirement at signup are not stated on UnoRouter's models or documentation pages.",
        "clock": "Standing best-effort catalog with a cap of 1 request per minute, per free model, per user, plus upstream provider caps, daily token budgets on some free pools that reset at midnight UTC, tokens-per-minute caps, and a per-user concurrency limit. No published expiry.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-47ceec7e955b38d12af6",
      "program_id": "zai-glm-flash-free",
      "event_type": "program_record_corrected",
      "program_name": "Z.ai / Zhipu free GLM model APIs",
      "provider_label": "Z.ai",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://docs.z.ai/guides/overview/pricing",
      "changed_fields": [
        "data_treatment",
        "clock"
      ],
      "before": {
        "data_treatment": "Z.AI's Additional Terms for API Services state that End User Content is used only as necessary to provide the API Services, comply with applicable law, enforce policies, and prevent abuse, and that it will not be used to develop or improve Services unless the user explicitly agrees. The same Terms of Use page states that for individual users Z.AI reserves the right to process any User Content to improve existing Services and develop new products and services, and that individual users authorize the use and storage of User Content that does not constitute personal data for developing and improving Z.AI's machine learning and artificial intelligence technologies. China-region BigModel terms separately allow anonymized information to be used for research, statistical analysis, service improvement, and machine learning or model-algorithm training.",
        "clock": "not_stated; no end date is attached to the models whose full input/output rows say Free"
      },
      "after": {
        "data_treatment": "Z.AI's Additional Terms for API Services state that End User Content is used only as necessary to provide the API Services, comply with applicable law, enforce policies, and prevent abuse, and that it will not be used to develop or improve Services unless the user explicitly agrees. The same Terms of Use say that for individual users Z.AI reserves the right to process User Content to improve existing Services and develop new products, and that individual users authorize the use and storage of User Content that is not personal data for developing and improving its machine learning and artificial intelligence technologies, so the protection depends on whether the account is treated as a business or developer account rather than an individual one. The China-region BigModel privacy policy separately allows anonymized information to be used for research, statistical analysis, service improvement, and machine learning or model-algorithm training.",
        "clock": "not_stated; no end date is attached to GLM-4.7-Flash or GLM-4.6V-Flash, the two models the international pricing page lists as Free on every price dimension."
      }
    },
    {
      "id": "free-program-1cd8a90a159a97a14a38",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On 2026-09-17, the pricing tables that mark standard Free-tier input and output as free of charge covered Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first Gemini Embedding, Gemini Robotics ER 1.6 Preview, and Gemini 2.5 Flash-Lite Preview no longer have pricing tables on that page. Access is model-specific and mode-specific: some models also mark Batch, Flex, or Priority modes free of charge on the Free tier, while other listed models and modes say Free tier 'Not available.'"
      },
      "after": {
        "models_detail": "Models whose pricing tables mark standard Free tier input and output as free of charge, checked on 2026-09-18, are Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemma 4, and two speech-to-text entries: Gemini 3.5 Transcribe Live for streaming transcription and Gemini 3.5 Transcribe for diarizing transcription. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding, Gemini 2.5 Flash-Lite Preview, and Gemini Robotics ER 1.6 Preview no longer appear on the pricing page on 2026-09-18. This is model-specific: other listed models, including Gemini 3.1 Pro Preview, the Nano Banana image models, Gemini Omni Flash, Veo 3.1, Lyria, and Gemini 2.5 Computer Use Preview, and many discounted or batch modes say Free tier 'Not available.'"
      }
    },
    {
      "id": "free-program-10ad9004fd1f51dd126f",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "models_detail",
        "signup"
      ],
      "before": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn via SMS or email login, API key from the console (SMS login favors Chinese numbers, email login supported)"
      },
      "after": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-18, several of them behind the page's expand buttons: XingChenAGI/Xing4.0-29B, THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, the ChinaTelecom XingChen speech models, Qwen/Qwen3-ASR-1.7B and FunAudioLLM/SenseVoiceSmall, and the Kwai-Kolors/Kolors image model. No Baidu ERNIE chat model is free.",
        "signup": "required; account at cloud.siliconflow.cn, where the platform supports login via SMS or email, with the API key created in the console (SMS login favors Chinese numbers, email login supported). The provider states that after verifying your identity you can use all free models, so identity verification is part of reaching the free roster."
      }
    },
    {
      "id": "free-program-075ac3b78c0fa01bb0d2",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-18T00:00:00+00:00",
      "date": "2026-09-18",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Eight catalog entries were listed at $0.00/1M on 2026-09-17: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Union Alpha; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and was listed at $0.11 in / $0.44 out per 1M on 2026-09-17."
      },
      "after": {
        "models_detail": "Seven catalog entries at $0.00/1M on 2026-09-18: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.11 in and $0.44 out per 1M."
      }
    },
    {
      "id": "free-event-df7387788399e9bc06c3",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "minimax/minimax-m2.7:free",
      "model": "MiniMax: MiniMax M2.7",
      "model_key": "minimaxm27free",
      "detected_at": "2026-09-17T15:28:04+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://openrouter.ai/minimax/minimax-m2.7:free"
    },
    {
      "id": "free-event-c07f365a5fafebe059b4",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "minimax/minimax-m3:free",
      "model": "MiniMax: MiniMax M3",
      "model_key": "minimaxm3free",
      "detected_at": "2026-09-17T15:28:04+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://openrouter.ai/minimax/minimax-m3:free"
    },
    {
      "id": "free-event-b81cc53b699ab9ca6694",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "inclusionai/ling-3.0-flash-vl:free",
      "model": "inclusionAI: Ling 3.0 Flash VL",
      "model_key": "ling30flashvlfree",
      "detected_at": "2026-09-17T15:28:04+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-vl:free"
    },
    {
      "id": "free-event-28c0e42ad7ac1a2aa400",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "nex-agi/nex-n2.5-pro:free",
      "model": "Nex AGI: Nex-N2.5-Pro",
      "model_key": "nexn25profree",
      "detected_at": "2026-09-17T15:28:04+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://openrouter.ai/nex-agi/nex-n2.5-pro:free"
    },
    {
      "id": "free-event-1f138b34d5de7befd62a",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "nex-agi/nex-n2.5-mini:free",
      "model": "Nex AGI: Nex-N2.5-Mini",
      "model_key": "nexn25minifree",
      "detected_at": "2026-09-17T15:28:04+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://openrouter.ai/nex-agi/nex-n2.5-mini:free"
    },
    {
      "id": "free-program-f510850551c4a64b8bd6",
      "program_id": "routelm-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "RouteLM monthly free tokens",
      "provider_label": "RouteLM",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://routelm.dev/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Model list not published; the free tokens apply across the routed catalog at provider rates."
      },
      "after": {
        "models_detail": "The 50,000 monthly free tokens apply only to the eight models the pricing table marks Eligible: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. The rest of the catalog (Claude, GPT, Grok, Kimi, MiniMax, GLM-5.2, DeepSeek V4 Pro) is marked Standard and is paid only."
      }
    },
    {
      "id": "free-program-efa0b3a91025d92cff5f",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Seven catalog entries are listed at $0.00 per 1M: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.10 in and $0.42 out per 1M."
      },
      "after": {
        "models_detail": "Eight catalog entries were listed at $0.00/1M on 2026-09-17: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Union Alpha; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and was listed at $0.11 in / $0.44 out per 1M on 2026-09-17."
      }
    },
    {
      "id": "free-program-ee99b5aacd0085bf1596",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_record_corrected",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day shared across the free catalog and reset every day. Models that carry a higher request weight have lower caps, published in the catalog table as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "56 models from 14 model authors listed at $0 input and $0 output on the provider's free catalog page (updated 2026-09-04), nearly all text models plus image-capable entries (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day, resetting daily (a few models are capped at 50 per day and 5 per minute).",
        "models_detail": "58 models listed at $0 input and $0 output on the provider's free catalog page (read 2026-09-17), nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time."
      }
    },
    {
      "id": "free-program-ea3cdab75550ffeb8be5",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_terms_changed",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Free hosted inference during the experiment. General rate limits exist and were reduced during the first week under load; requests can queue at peak. Exact numeric limits are not published on a publicly readable surface; the docs page at experiments.hetzner.com sits behind Hetzner's own CAPTCHA.",
        "models_detail": "Currently Qwen3.6-35B-A3B and Qwen 3.8 (small models). GLM-5.2, DeepSeek-V4-Flash-0731, and Kimi-K2.7-Code were served at launch, but Hetzner announced a break from large models in the 2026-08-17 blog post.",
        "card": "not_stated",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will email in advance if that changes. Published limits per API key: 10 requests per 60 seconds, 4M input tokens and 100K output tokens per 60 seconds, HTTP 429 past any of them. Performance and availability are not guaranteed, requests can queue at peak, and the platform is not for production use.",
        "models_detail": "On the Inference API docs page (2026-09-17): Qwen/Qwen3.6-35B-A3B-FP8 and Qwen3.8-27B, both 262,144-token context with text and image input. The models endpoint is the definitive list and the selection changes during the experiment; GLM-5.2, DeepSeek-V4-Flash-0731 and Kimi-K2.7-Code were served at launch and dropped when Hetzner paused large models in August 2026.",
        "card": "not required: Hetzner's account-creation guide lists an email address, contact details and a billing address as the requirements, with adding a payment method only an optional later step",
        "card_required": "no"
      }
    },
    {
      "id": "free-program-e3aab144408f08fae694",
      "program_id": "nvidia-nim-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "160+ NIM API endpoints (Developer Program)",
      "provider_label": "NVIDIA",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://developer.nvidia.com/nim",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership; the NVIDIA NIM for Developers page describes free access to NIM API endpoints for unlimited prototyping, powered by DGX Cloud. Production use is excluded and requires an NVIDIA AI Enterprise license. NVIDIA's NIM FAQ describes the API catalog as a trial experience whose number of allowed calls is designed to serve evaluation and prototyping needs only, with the rate of requests varying per model queried and varying with the number of concurrent users; exact rate limits are shown in the authenticated account at build.nvidia.com. Developer Program membership also licenses self-hosted NIM microservices for research, application development, and experimentation on up to 16 GPUs."
      },
      "after": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership. Production use is excluded. NVIDIA's NIM FAQ says the API catalog is a trial experience, that the rate of requests varies per model and with concurrent users, and that the exact limits are shown in the authenticated account; no public page states a number (the build.nvidia.com FAQ that once said 40 requests per minute now returns 404). Developer Program membership also licenses self-hosted NIM microservices for non-production use on up to 16 GPUs."
      }
    },
    {
      "id": "free-program-df1ab1732e3c1a5aa3ce",
      "program_id": "catalog2-orcarouter",
      "event_type": "program_terms_changed",
      "program_name": "OrcaRouter free models",
      "provider_label": "OrcaRouter",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://docs.orcarouter.ai/routing/free-models",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "clock"
      ],
      "before": {
        "free_terms": "OrcaRouter provides orcarouter/free and explicit -free model IDs that never touch the account wallet and can be used indefinitely, subject to unpublished, dynamically tuned request-rate, daily, and prompt-size limits. The general orcarouter/auto router can resolve to ordinary paid models billed at upstream prices, so it is not part of this free program.",
        "models_detail": "orcarouter/free plus the explicit -free model IDs. The orcarouter/auto alias can select paid models and is not part of this free program.",
        "signup": "OrcaRouter account and API key required; free routes never touch the wallet.",
        "clock": "Standing dynamic free route with unpublished, automatically tuned request, daily, and prompt-size limits."
      },
      "after": {
        "free_terms": "OrcaRouter provides orcarouter/free and explicit -free model IDs that never touch the account wallet and can be called indefinitely within rate limits. Requests are capped per minute and per day, and the day window resets at 00:00 UTC. The rpm and rpd numbers for each tier are published at GET /api/free-package/public and are tuned live. Limits scale with the account's lifetime spend: an account that has never topped up gets a small daily allowance. Lower tiers also have a per-request prompt-size cap, and that number is not published. The general orcarouter/auto router can resolve to ordinary paid models billed at upstream prices, so it is not part of this free program.",
        "models_detail": "orcarouter/free plus the explicit -free model IDs, which rotate as capacity allows. On 2026-09-17, orcarouter/free listed one model: deepseek/deepseek-v4-flash-free. The orcarouter/auto alias can select paid models and is not part of this free program.",
        "signup": "An OrcaRouter account and API key are required. Free requests can be refused unless the workspace owner links a GitHub account that has been registered for some time, or the workspace has paid enough to be exempt. Free routes never touch the wallet.",
        "clock": "Standing free route. Per-minute and per-day request limits are published per tier and tuned live, and the per-request prompt-size cap is unpublished."
      }
    },
    {
      "id": "free-program-d887103124baba948231",
      "program_id": "routelm-free-tier",
      "event_type": "program_record_corrected",
      "program_name": "RouteLM monthly free tokens",
      "provider_label": "RouteLM",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://routelm.dev/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Model list not published; the free tokens apply across the routed catalog at provider rates."
      },
      "after": {
        "models_detail": "The 50,000 monthly free tokens apply only to the eight models the pricing table marks Eligible: deepseek-v4-flash, deepseek-v4-flash-0731, glm-4.7-flash, gpt-oss-120b, gpt-oss-20b, laguna-s-2.1, laguna-xs-2.1, and qwen3.5-9b. The rest of the catalog (Claude, GPT, Grok, Kimi, MiniMax, GLM-5.2, DeepSeek V4 Pro) is marked Standard and is paid only."
      }
    },
    {
      "id": "free-program-d2f08d1c7991a08d79aa",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. The catalog contains over 40 pre-trained models and includes Llama, Qwen, DeepSeek, GPT-OSS, embedding, image-generation, and speech models. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalog also prices a few entries at Free outright, on 2026-09-04 the beta Qwen3Guard-Gen 0.6B and 8B moderation models (Qwen3.8-27B itself is paid at 0.4 EUR in / 2.7 EUR out per 1M tokens)."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model. OVHcloud's product page describes over 40 models, such as Llama, Qwen and DeepSeek, but the public catalog listed 20 models on 2026-09-17: Qwen3.8-27B, Qwen3.6-27B, Qwen3.5-397B-A17B, Qwen3.5-9B, Qwen3-Embedding-8B, Qwen3Guard-Gen-8B, Qwen3Guard-Gen-0.6B, gpt-oss-120b, gpt-oss-20b, whisper-large-v3, whisper-large-v3-turbo, Meta-Llama-3_3-70B-Instruct, Qwen2.5-VL-72B-Instruct, bge-multilingual-gemma2, bge-m3, stable-diffusion-xl-base-v10, and four NVIDIA Riva text-to-speech voices. Models announced for retirement stay callable through the API but are hidden from the catalog. The authenticated catalog also prices a few entries at Free outright, on 2026-09-17 the beta Qwen3Guard-Gen 0.6B and 8B moderation models (Qwen3.8-27B itself is paid at 0.4 EUR in / 2.7 EUR out per 1M tokens)."
      }
    },
    {
      "id": "free-program-b9ac5f100fd34793fd43",
      "program_id": "mancer-free-models",
      "event_type": "program_terms_changed",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "MythoLite is the only model on Mancer's public model list priced FREE for both input and output tokens. It is listed as a LLaMA 2 13B model tagged for creative and roleplay use, and it was shown as online on 2026-09-05."
      },
      "after": {
        "models_detail": "MythoLite only: a LLaMA 2 13B creative/roleplay model priced FREE for input and output on the current model table. Its context and output caps sit behind a details fold the nightly read does not expand, so they are not stated here."
      }
    },
    {
      "id": "free-program-a9a543d1f2aef7ecf01c",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "On the Gemini Developer API pricing page as of 2026-09-04, models whose standard Free-tier input and output are marked \"Free of charge\" include Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash Live Preview, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio (Live API), Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Embedding, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemini Robotics ER 1.6 Preview, and Gemma 4. Free-tier availability is model-specific: other listed models, such as Gemini 3.1 Pro Preview, the Nano Banana image models, Gemini Omni Flash, Veo 3.1, Lyria, and Gemini 2.5 Computer Use Preview, show Free tier \"Not available,\" and some batch or discounted modes may also show \"Not available.\" No pricing table for Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, or Gemini 2.5 Flash-Lite Preview appears on the pricing page."
      },
      "after": {
        "models_detail": "On 2026-09-17, the pricing tables that mark standard Free-tier input and output as free of charge covered Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.8 Live, Gemini 3.8 Live Extended Thinking, Gemini 3.1 Flash Live Preview, Gemini 3.5 Live Translate, Gemini 3.5 Transcribe Live, Gemini 3.5 Transcribe, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini Embedding 2, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, and Gemma 4. Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, the first Gemini Embedding, Gemini Robotics ER 1.6 Preview, and Gemini 2.5 Flash-Lite Preview no longer have pricing tables on that page. Access is model-specific and mode-specific: some models also mark Batch, Flex, or Priority modes free of charge on the Free tier, while other listed models and modes say Free tier 'Not available.'"
      }
    },
    {
      "id": "free-program-a646b92fe0022a63b5eb",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "card_required"
      ],
      "before": {
        "free_terms": "Kenari currently offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (11 models on 2026-09-04) at up to 5 requests per minute. Its quickstart explicitly says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort and has no speed or availability guarantee.",
        "models_detail": "Eleven provider-hosted :free text models per Kenari's homepage counter on 2026-09-04 (it read ten the day before and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "signup": "Free Kenari account and API key required; users may skip top-up and no card is required for the Solo free lane.",
        "card_required": "no"
      },
      "after": {
        "free_terms": "Kenari offers free account registration and a Solo tier with no top-up required: 50 requests per day across its free models (17 models on 2026-09-17) at up to 5 requests per minute. Its quickstart says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort with no speed or availability guarantee, and Kenari says it can change the quota and rate limit values at any time.",
        "models_detail": "Seventeen :free models per the Solo tier on Kenari's homepage on 2026-09-17 (it read eleven on 2026-09-04 and twelve on 2026-08-27; the roster moves with the provider catalog).",
        "signup": "A free Kenari account and API key are required, and users may skip top-up to use :free models. Kenari's quickstart says no foreign credit card is needed; it does not state whether any card is required.",
        "card_required": "not_stated"
      }
    },
    {
      "id": "free-program-912b665a59f251510225",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "10 requests per minute and 50 requests per day on current free-quota models; accounts that have topped up get twice the daily quota. Past the quota, requests continue at the paid rate only when the account holds credit, otherwise they are rate-limited until the quota resets.",
        "models_detail": "One current free-quota model (Qwen: Qwen3.7 Flash) plus auto:free routing; the catalog changes over time."
      },
      "after": {
        "free_terms": "10 requests per minute and 50 requests per day on each current free-quota model, with the daily quota counted separately per model; accounts that have topped up get twice the daily quota. The provider's documentation says the daily quota is measured in units that grow with prompt context length, so long-context requests use it up faster, and auto:free routing has its own separate per-IP limit. Past the quota, requests continue at the paid rate only when the account holds credit, otherwise they are rate-limited until the quota resets.",
        "models_detail": "Two current free-quota models on 2026-09-17 (Qwen: Qwen3.7 Flash and DeepSeek: Deepseek V4 Flash 0731free) plus auto:free routing; the catalog changes over time."
      }
    },
    {
      "id": "free-program-80d8c0343ff7b721b325",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_record_corrected",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "not_stated; a positive account balance is required, but the reviewed offer page does not require that balance to be funded specifically by a card"
      },
      "after": {
        "card": "not stated as a card; a positive prepaid balance is required to use the free tokens, and OpenAI's prepaid-billing article puts the minimum initial credit purchase at $5 (payment details are added in that step). The free daily tokens are not billed, so they never draw the balance down."
      }
    },
    {
      "id": "free-program-7e39a9fab15db3a7889b",
      "program_id": "openai-data-share-tokens",
      "event_type": "program_terms_changed",
      "program_name": "Free daily tokens for opted-in data sharing",
      "provider_label": "OpenAI",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://help.openai.com/en/articles/10306912-sharing-feedback-evaluation-and-fine-tuning-data-and-api-inputs-and-outputs-with-openai",
      "changed_fields": [
        "card"
      ],
      "before": {
        "card": "not_stated; a positive account balance is required, but the reviewed offer page does not require that balance to be funded specifically by a card"
      },
      "after": {
        "card": "not stated as a card; a positive prepaid balance is required to use the free tokens, and OpenAI's prepaid-billing article puts the minimum initial credit purchase at $5 (payment details are added in that step). The free daily tokens are not billed, so they never draw the balance down."
      }
    },
    {
      "id": "free-program-7d081ea12ff644b37b47",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "card"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day shared across the free catalog and reset every day. Models that carry a higher request weight have lower caps, published in the catalog table as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "56 models from 14 model authors listed at $0 input and $0 output on the provider's free catalog page (updated 2026-09-04), nearly all text models plus image-capable entries (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "signup": "Account required; no credit card for the initial ten calls. A one-time $1 top-up unlocks the daily quotas.",
        "card": "not required for the 10 trial calls; a one-time $1 top-up (not a card on file) unlocks the daily free quotas"
      },
      "after": {
        "free_terms": "Every account starts with 10 trial calls shared across all free models, with no credit card and no expiry. A one-time top-up of $1 or more permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 per minute, and 1M tokens per day shared across all free models, reset daily. Some models carry a higher request weight and have lower caps, including 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "58 models listed at $0 input and $0 output on the provider's free catalog page on 2026-09-16, nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time.",
        "signup": "Account required, created by email or OAuth; no credit card for the initial ten calls. A one-time top-up of $1 or more unlocks the daily quotas.",
        "card": "Not required for the 10 trial calls. The daily free quotas require a one-time top-up of $1 or more; the provider does not state which payment methods a top-up accepts."
      }
    },
    {
      "id": "free-program-72c661e4cff010ea4267",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Six free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "models_detail",
        "data_treatment",
        "training_use"
      ],
      "before": {
        "models_detail": "Free rows on the Zen pricing table on 2026-09-05: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, and Muse Spark 1.3 Contributor Free. Muse Spark 1.2 Contributor Free no longer appears on the pricing table, the endpoints table, or the free-model notes, and Muse Spark 1.2 is listed at $1.25 per million input tokens and $4.25 per million output tokens. Ox Alpha Free and Hy3 Free were removed from the free table before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models.",
        "training_use": "yes"
      },
      "after": {
        "models_detail": "Free rows on the Zen pricing table on 2026-09-17: Big Pickle, Union Alpha Free, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, and Muse Spark 1.3 Contributor Free. Union Alpha Free was added and Muse Spark 1.2 Contributor Free was removed from the free table by 2026-09-17. Ox Alpha Free and Hy3 Free were removed before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models. Union Alpha Free's provider follows a zero-retention policy and does not use your data for model training.",
        "training_use": "varies"
      }
    },
    {
      "id": "free-program-6c2d8f68933169ad5ec4",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-04: NVIDIA Nemotron 3 Ultra 550B, Nemotron 3 Super 120B, Nemotron 3 Nano 30B, Nemotron 3 Nano Omni 30B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that omits Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty states that it never uses customer data for training, and its model FAQ says each model detail page shows the specific data retention and training policy for the provider serving that model. Because free models run through different underlying providers, check the relevant model and provider policy before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-17: NVIDIA Nemotron 3.5 Lightning 30B A3B, Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, and Nemotron 3.5 Content Safety; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; InclusionAI Ling 3.0 Tiny; and Meta Muse Glimmer 30B. The docs page carries a shorter nine-row table. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "No credit card required for the Free plan.",
        "data_treatment": "Requesty says it never uses your data for training. Data retention and training policy for each model depends on the underlying provider and is shown on that model's detail page, so check it before sensitive use."
      }
    },
    {
      "id": "free-program-3a5325623db853937c67",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "signup",
        "card",
        "data_treatment",
        "clock",
        "card_required",
        "access_clock"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. Amazon's Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is \"Free of charge with daily limits\"; the alternative AWS IAM authentication path instead applies usage charges to an AWS account. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). No numeric quota or rate-limit figures are published publicly.",
        "models_detail": "Amazon Nova family on the nova.amazon.com developer surface; page metadata names Amazon Nova 2 Lite, Nova Act, and Nova 2 Sonic. The per-model roster is behind Amazon sign-in and is not publicly confirmable.",
        "signup": "required; Amazon.com consumer account sign-in, then an API key from the dev surface",
        "card": "not required; no payment terms exist anywhere in the Nova Terms of Use, and the Nova Act user guide describes API key authentication as free of charge with daily limits",
        "data_treatment": "Interactions are recorded, retained, may be manually reviewed, and are used to provide, develop, and improve Amazon's services including artificial intelligence models (Terms 1.2). No opt-out is stated.",
        "clock": "Standing developer offering (live since March 2025, terms current as of Dec 2025); no end date stated.",
        "card_required": "no",
        "access_clock": "standing"
      },
      "after": {
        "free_terms": "The Amazon Nova Act developer guide says API key access from nova.amazon.com/act is free of charge with daily limits. The daily limit numbers are not published on these pages. Under the Nova.amazon.com Terms of Use (last updated December 2, 2025), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and some software requires an Amazon-provided API key (Terms 1.5). The Build with Nova schedule describes an environment for experimentation that does not support deploying production-scale developer offerings (S-1.1). For production use, AWS IAM authentication is recommended, and usage charges apply to the AWS account.",
        "models_detail": "Amazon Nova models and capabilities on nova.amazon.com. The Nova Act developer guide names Nova Act as reachable with a free API key. On 2026-09-17, the public pages did not publish a per-model roster.",
        "signup": "Required: the API key comes from nova.amazon.com/act and is tied to an Amazon.com account.",
        "card": "Not stated by the provider.",
        "data_treatment": "Amazon records, processes and retains interactions to provide, develop and improve its services, including artificial intelligence models, and may review them manually (Terms 1.2). Information may be stored on servers outside the user's country (Terms 2.1). No opt-out is stated.",
        "clock": "No end date is stated. Amazon may change, suspend or discontinue Amazon Nova at any time without notice (Terms 2.3).",
        "card_required": "not_stated",
        "access_clock": "not_stated"
      }
    },
    {
      "id": "free-program-34b172429c2a6ba3dcef",
      "program_id": "catalog2-inferx",
      "event_type": "program_delisted",
      "program_name": "InferX Free plan (Qwen3.6 models)",
      "provider_label": "InferX",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://inferx.net/pricing",
      "delisted_status": "uncertain"
    },
    {
      "id": "free-program-2fbcc209af4fe958269f",
      "program_id": "nvidia-nim-dev-free",
      "event_type": "program_record_corrected",
      "program_name": "160+ NIM API endpoints (Developer Program)",
      "provider_label": "NVIDIA",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://developer.nvidia.com/nim",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership; the NVIDIA NIM for Developers page describes free access to NIM API endpoints for unlimited prototyping, powered by DGX Cloud. Production use is excluded and requires an NVIDIA AI Enterprise license. NVIDIA's NIM FAQ describes the API catalog as a trial experience whose number of allowed calls is designed to serve evaluation and prototyping needs only, with the rate of requests varying per model queried and varying with the number of concurrent users; exact rate limits are shown in the authenticated account at build.nvidia.com. Developer Program membership also licenses self-hosted NIM microservices for research, application development, and experimentation on up to 16 GPUs."
      },
      "after": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership. Production use is excluded. NVIDIA's NIM FAQ says the API catalog is a trial experience, that the rate of requests varies per model and with concurrent users, and that the exact limits are shown in the authenticated account; no public page states a number (the build.nvidia.com FAQ that once said 40 requests per minute now returns 404). Developer Program membership also licenses self-hosted NIM microservices for non-production use on up to 16 GPUs."
      }
    },
    {
      "id": "free-program-283a14f0232b7e51bcb6",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash and 3.0 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash at $0 per 1M input and output tokens; agnes-2.0-flash is deprecated and kept on the page as a historical record. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to about 20 RPM. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "agnes-2.5-flash is currently listed at $0 input and $0 output; agnes-2.0-flash is marked deprecated, with its $0 prices kept only as historical records."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash and agnes-3.0-flash at $0 per 1M input, output and cached-input tokens; agnes-2.0-flash is deprecated and kept on the page as a historical record. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to about 20 RPM. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "agnes-2.5-flash and agnes-3.0-flash are listed at $0 input, $0 output and $0 cached input (the pricing page says all three are currently free for both models); agnes-2.0-flash is marked deprecated, with its $0 prices kept only as historical records."
      }
    },
    {
      "id": "free-program-1a3a693619939bf0587e",
      "program_id": "hetzner-inference-experiment",
      "event_type": "program_record_corrected",
      "program_name": "Hetzner Inference Experiment",
      "provider_label": "Hetzner",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://www.hetzner.com/blog/inference-experiment/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card",
        "card_required"
      ],
      "before": {
        "free_terms": "Free hosted inference during the experiment. General rate limits exist and were reduced during the first week under load; requests can queue at peak. Exact numeric limits are not published on a publicly readable surface; the docs page at experiments.hetzner.com sits behind Hetzner's own CAPTCHA.",
        "models_detail": "Currently Qwen3.6-35B-A3B and Qwen 3.8 (small models). GLM-5.2, DeepSeek-V4-Flash-0731, and Kimi-K2.7-Code were served at launch, but Hetzner announced a break from large models in the 2026-08-17 blog post.",
        "card": "not_stated",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Free hosted inference while the Inference API is in experimental status; Hetzner says it will email in advance if that changes. Published limits per API key: 10 requests per 60 seconds, 4M input tokens and 100K output tokens per 60 seconds, HTTP 429 past any of them. Performance and availability are not guaranteed, requests can queue at peak, and the platform is not for production use.",
        "models_detail": "On the Inference API docs page (2026-09-17): Qwen/Qwen3.6-35B-A3B-FP8 and Qwen3.8-27B, both 262,144-token context with text and image input. The models endpoint is the definitive list and the selection changes during the experiment; GLM-5.2, DeepSeek-V4-Flash-0731 and Kimi-K2.7-Code were served at launch and dropped when Hetzner paused large models in August 2026.",
        "card": "not required: Hetzner's account-creation guide lists an email address, contact details and a billing address as the requirements, with adding a payment method only an optional later step",
        "card_required": "no"
      }
    },
    {
      "id": "free-program-048312a28afced16ad89",
      "program_id": "mancer-free-models",
      "event_type": "program_record_corrected",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-17T00:00:00+00:00",
      "date": "2026-09-17",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "MythoLite is the only model on Mancer's public model list priced FREE for both input and output tokens. It is listed as a LLaMA 2 13B model tagged for creative and roleplay use, and it was shown as online on 2026-09-05."
      },
      "after": {
        "models_detail": "MythoLite only: a LLaMA 2 13B creative/roleplay model priced FREE for input and output on the current model table. Its context and output caps sit behind a details fold the nightly read does not expand, so they are not stated here."
      }
    },
    {
      "id": "free-event-77252451ba5f805ed356",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "inclusionai/ling-3.0-flash-sante:free",
      "model": "inclusionAI: Ling 3.0 Flash Sante",
      "model_key": "ling30flashsantefree",
      "detected_at": "2026-09-05T11:30:10+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-sante:free"
    },
    {
      "id": "free-program-e671d06481e521e0ba27",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Six free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-05T00:00:00+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment",
        "clock"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices seven models at Free for input, output, and cached-read tokens. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the current Zen pricing table: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free (added 2026-09-04), and Muse Spark 1.2 Contributor Free. Ox Alpha Free and Hy3 Free were removed from the free table before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free and Muse Spark 1.2 Contributor Free permit prompts and completions to train future Meta models.",
        "clock": "All seven are limited-time promotions; no end date is stated for any of them."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices six models at Free for input, output, and cached-read tokens. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the Zen pricing table on 2026-09-05: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, and Muse Spark 1.3 Contributor Free. Muse Spark 1.2 Contributor Free no longer appears on the pricing table, the endpoints table, or the free-model notes, and Muse Spark 1.2 is listed at $1.25 per million input tokens and $4.25 per million output tokens. Ox Alpha Free and Hy3 Free were removed from the free table before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free permits prompts and completions to train future Meta models.",
        "clock": "All six are limited-time promotions; no end date is stated for any of them."
      }
    },
    {
      "id": "free-program-d9724deafe7aa29eb8ce",
      "program_id": "mancer-free-models",
      "event_type": "program_terms_changed",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-05T00:00:00+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "MythoLite only: on 2026-09-04 the Mancer model table listed MythoLite as a LLaMA 2 13B model tagged Creative and Roleplay, with both input and output token cost shown as FREE. The model table and model detail panel do not currently display a context length or maximum output figure for MythoLite."
      },
      "after": {
        "models_detail": "MythoLite is the only model on Mancer's public model list priced FREE for both input and output tokens. It is listed as a LLaMA 2 13B model tagged for creative and roleplay use, and it was shown as online on 2026-09-05."
      }
    },
    {
      "id": "free-program-48463d3d64119e4ef2f9",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-05T00:00:00+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail",
        "card",
        "data_treatment"
      ],
      "before": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-04: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, and Nemotron 3.5 Content Safety; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; Muse Glimmer 30B; and one further open model. The docs page carries a shorter nine-row table. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "not_stated",
        "data_treatment": "Not stated uniformly because routes can involve different model providers; check Requesty and underlying provider policies before sensitive use."
      },
      "after": {
        "models_detail": "Requesty's free models page listed 12 free models on 2026-09-04: NVIDIA Nemotron 3 Ultra 550B, Nemotron 3 Super 120B, Nemotron 3 Nano 30B, Nemotron 3 Nano Omni 30B Reasoning, Nemotron 3.5 Content Safety and Nemotron 3.5 Lightning 30B; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; and Muse Glimmer 30B. The documentation page carries a shorter nine-row table that omits Nemotron 3.5 Lightning, Ling 3.0 Tiny and Muse Glimmer. Requesty says the set is free for now and announces changes in its changelog.",
        "card": "Requesty's pricing page states that the Free plan requires no credit card.",
        "data_treatment": "Requesty states that it never uses customer data for training, and its model FAQ says each model detail page shows the specific data retention and training policy for the provider serving that model. Because free models run through different underlying providers, check the relevant model and provider policy before sensitive use."
      }
    },
    {
      "id": "free-program-24d6ed14a517490b76d9",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-05T00:00:00+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The published catalog currently lists seven entries priced at $0.00 per 1M tokens: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; and Upstage Solar Pro 4. Tencent Hy3 is no longer part of the free set and is listed at $0.07 in / $0.26 out per 1M. The free set changes over time as the catalog is updated."
      },
      "after": {
        "models_detail": "Seven catalog entries are listed at $0.00 per 1M: inclusionAI Ling 3.0 Flash Fin; inclusionAI Ling 3.0 Flash Sante (free); Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 is not in the free set and is listed at $0.10 in and $0.42 out per 1M."
      }
    },
    {
      "id": "free-program-0e8c5a72cf94a9fd9866",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-05T00:00:00+00:00",
      "date": "2026-09-05",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across the whole free catalog; when they are used up the API keeps answering with a note that the trial is exhausted. A one-time top-up of any amount from $1 permanently switches the account to daily quotas on the free catalog: 100 requests per day, 10 requests per minute and 1M tokens per day, shared across all free models and reset every day. Models that carry a higher request weight have lower per-model caps, 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute on gpt-4o-free and gpt-4.1-free, and 10 per day and 1 per minute on gemini-3.1-flash-image-preview-free. Past a cap the API returns HTTP 429 until the next minute or day begins.",
        "models_detail": "56 models from 14 model authors listed at $0 input and $0 output on the provider's free catalog page, which was marked updated on 2026-09-04; nearly all are text models, alongside a few image-capable entries (gpt-image-2-free, gemini-3.1-flash-image-preview-free). The roster changes over time."
      },
      "after": {
        "free_terms": "Every new account receives 10 trial calls with no credit card and no expiry, shared across all free models. A one-time top-up of at least $1 switches the account to permanent daily quotas on the free catalog: 100 requests per day, 10 requests per minute, and 1M tokens per day shared across the free catalog and reset every day. Models that carry a higher request weight have lower caps, published in the catalog table as 50 per day and 5 per minute, 33 per day and 3 per minute, 20 per day and 2 per minute, and 10 per day and 1 per minute for the image preview model. Past a cap the API returns 429 until the next minute or day starts.",
        "models_detail": "56 models from 14 model authors listed at $0 input and $0 output on the provider's free catalog page (updated 2026-09-04), nearly all text models plus image-capable entries (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time."
      }
    },
    {
      "id": "free-program-f6d5aff61dd2c61755f5",
      "program_id": "catalog2-kenari",
      "event_type": "program_terms_changed",
      "program_name": "Kenari Solo free lane",
      "provider_label": "Kenari",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://kenari.id/",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Kenari currently offers free account registration and a Solo tier requiring no card, with 50 requests per day across 12 provider-hosted free models at up to 5 requests per minute. Its quickstart explicitly says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort and has no speed or availability guarantee.",
        "models_detail": "Twelve dynamic provider-hosted :free text models in Kenari's current catalog."
      },
      "after": {
        "free_terms": "Kenari currently offers free account registration and a Solo tier requiring no card, with 50 requests per day across the current free-model roster (11 models on 2026-09-04) at up to 5 requests per minute. Its quickstart explicitly says users may skip top-up, create a Kenari API key, and call a :free text model through the OpenAI-compatible /v1/chat/completions endpoint without reducing their balance. The free lane is best-effort and has no speed or availability guarantee.",
        "models_detail": "Eleven provider-hosted :free text models per Kenari's homepage counter on 2026-09-04 (it read ten the day before and twelve on 2026-08-27; the roster moves with the provider catalog)."
      }
    },
    {
      "id": "free-program-ee78f7eaa1c4392bf3b7",
      "program_id": "cloudflare-workers-ai-free",
      "event_type": "program_terms_changed",
      "program_name": "Workers AI catalog (neuron-metered)",
      "provider_label": "Cloudflare",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://developers.cloudflare.com/workers-ai/platform/pricing/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The free Neuron allocation is offer-wide across the Workers AI catalog except models that explicitly require paid billing. The current exclusion list is @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, @cf/zai-org/glm-5.2, @cf/deepseek-ai/deepseek-v4-flash-0731, and @cf/deepseek-ai/deepseek-v4-pro-0813. Model-specific Neuron consumption varies by model and input/output type."
      },
      "after": {
        "models_detail": "The free Neuron allocation is offer-wide across the Workers AI catalog except models that explicitly require paid billing. The current exclusion list is @cf/moonshotai/kimi-k2.6, @cf/moonshotai/kimi-k2.7-code, @cf/zai-org/glm-5.2, @cf/zai-org/glm-5.3, @cf/zai-org/glm-5.3-flash, @cf/deepseek-ai/deepseek-v4-flash-0731, and @cf/deepseek-ai/deepseek-v4-pro-0813 (as of 2026-09-04); those models can be reached with Workers Paid or prepaid AI Gateway credits. Model-specific Neuron consumption varies by model and input/output type."
      }
    },
    {
      "id": "free-program-ecfd84d141ce94e91e1b",
      "program_id": "zai-glm-flash-free",
      "event_type": "program_terms_changed",
      "program_name": "Z.ai / Zhipu free GLM model APIs",
      "provider_label": "Z.ai",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://docs.z.ai/guides/overview/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Standing model-specific zero price: input, cached input, cached-input storage, and output are all listed as Free for GLM-4.7-Flash, GLM-4.5-Flash, and GLM-4.6V-Flash. Numeric request/token rate limits are not public on the pricing page and must be checked on the authenticated account rate-limits surface.",
        "models_detail": "GLM-4.7-Flash and GLM-4.5-Flash are the current free text models; GLM-4.6V-Flash is the current free vision-language model. Other models' 'Limited-time Free' cached-input storage is not free inference and is not included in this offer. China-region BigModel additionally lists GLM-4-Flash-250414, GLM-4.1V-Thinking-Flash, GLM-4V-Flash, CogView-3-Flash, and CogVideoX-Flash alongside overlapping GLM Flash models."
      },
      "after": {
        "free_terms": "Standing model-specific zero price: input, cached input, cached-input storage, and output are all listed as Free for GLM-4.7-Flash and GLM-4.6V-Flash (and the legacy GLM-4.5-Flash id, which BigModel says now routes to GLM-4.7-Flash). Numeric request/token rate limits are not public on the pricing page and must be checked on the authenticated account rate-limits surface.",
        "models_detail": "GLM-4.7-Flash is the current free text model and GLM-4.6V-Flash the current free vision-language model on the international pricing page. That page still prints GLM-4.5-Flash at Free on every price dimension, but the China-region BigModel page for GLM-4.5-Flash states the model was retired on 2026-01-30 with requests auto-routed to GLM-4.7-Flash, so treat GLM-4.5-Flash as a legacy id rather than a second free model. Other models' 'Limited-time Free' cached-input storage is not free inference and is not included in this offer. China-region BigModel additionally lists GLM-4-Flash-250414, GLM-4.1V-Thinking-Flash, GLM-4V-Flash, CogView-3-Flash, and CogVideoX-Flash alongside overlapping GLM Flash models."
      }
    },
    {
      "id": "free-program-eaf056d4920ac5f5614e",
      "program_id": "bazaarlink-free-models",
      "event_type": "program_terms_changed",
      "program_name": "BazaarLink free-model gateway",
      "provider_label": "BazaarLink",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://bazaarlink.ai/free",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "10 RPM and 50 requests/day on current free-quota models.",
        "models_detail": "One current zero-cost model plus auto:free routing; catalog changes over time."
      },
      "after": {
        "free_terms": "10 requests per minute and 50 requests per day on current free-quota models; accounts that have topped up get twice the daily quota. Past the quota, requests continue at the paid rate only when the account holds credit, otherwise they are rate-limited until the quota resets.",
        "models_detail": "One current free-quota model (Qwen: Qwen3.7 Flash) plus auto:free routing; the catalog changes over time."
      }
    },
    {
      "id": "free-program-e0410247706f445989f2",
      "program_id": "nous-portal-free-plan",
      "event_type": "program_terms_changed",
      "program_name": "Nous Portal Free plan",
      "provider_label": "Nous Portal",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://portal.nousresearch.com/",
      "changed_fields": [
        "models_detail",
        "data_treatment",
        "training_use"
      ],
      "before": {
        "models_detail": "Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Tencent Hy3; Upstage Solar Pro 4.",
        "data_treatment": "Nous terms describe safeguards for account data but the public plan page does not state whether free inference prompts are used for training; treat as not publicly stated.",
        "training_use": "not_stated"
      },
      "after": {
        "models_detail": "Six catalog entries at $0.00/1M on 2026-09-04: inclusionAI Ling 3.0 Flash Fin; Meituan LongCat 2.0; Poolside Laguna S 2.1; Poolside Laguna XS 2.1; StepFun Step 3.7 Flash; Upstage Solar Pro 4. Tencent Hy3 left the free set (now $0.11 in / $0.42 out per 1M).",
        "data_treatment": "Nous Portal's terms of service say Nous Research may use Your Data for research and for developing, training, fine-tuning, and improving AI and machine-learning models, under a broad license to process it; the plan page itself says nothing more specific about free-inference prompts.",
        "training_use": "yes"
      }
    },
    {
      "id": "free-program-dab8b619d6ef83ce1466",
      "program_id": "nvidia-nim-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "160+ NIM API endpoints (Developer Program)",
      "provider_label": "NVIDIA",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://developer.nvidia.com/nim",
      "changed_fields": [
        "free_terms"
      ],
      "before": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership. Production use is excluded. The build.nvidia.com FAQ states the free tier runs on rate limits of up to 40 requests per minute for most models with no per-token billing; exact limits are shown in the authenticated account. Developer Program membership also licenses self-hosted NIM microservices for non-production use on up to 16 GPUs."
      },
      "after": {
        "free_terms": "Free NVIDIA-hosted NIM API endpoints for prototyping, research, development, testing, learning, and evaluation through free NVIDIA Developer Program membership. Production use is excluded. The build.nvidia.com FAQ states the free tier runs on rate limits of up to 40 requests per minute for most models with no per-token billing, and NVIDIA's NIM FAQ adds that the rate varies per model and with concurrent users; exact limits are shown in the authenticated account. Developer Program membership also licenses self-hosted NIM microservices for non-production use on up to 16 GPUs."
      }
    },
    {
      "id": "free-program-daa44132648c3a642161",
      "program_id": "catalog2-meganova",
      "event_type": "program_terms_changed",
      "program_name": "MegaNova free-quota model (Mistral Small 3.2)",
      "provider_label": "Meganova",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.meganova.ai/mistralai/Mistral-Small-3.2-24B-Instruct-2506",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "card"
      ],
      "before": {
        "free_terms": "MegaNova hosts `mistralai/Mistral-Small-3.2-24B-Instruct-2506` as a recurring free-quota text model. Free Tier 1 registration requires no credit card and includes 50 requests per day for this model; the documented limits rise to 300, 300, and 1,000 requests per day across Tiers 2\u20134. Quotas reset daily at 00:00 UTC. After exhaustion, calls fail unless the user has enabled the optional paid fallback, so this is genuine standing free hosted inference but not unlimited `$0` usage.",
        "models_detail": "mistralai/Mistral-Small-3.2-24B-Instruct-2506 is the documented recurring free-quota text model.",
        "card": "not_stated"
      },
      "after": {
        "free_terms": "MegaNova hosts `mistralai/Mistral-Small-3.2-24B-Instruct-2506` as a recurring free-quota text model. Free Tier 1 registration requires no credit card and includes 50 requests per day for this model, within account rate limits of 60 requests per minute and 200,000 tokens per minute. Free quota is granted per account and resets daily at 00:00 UTC. Documentation states that Tier 2 and above keep free-quota access with higher limits, but the published quota table does not list per-tier request numbers for this model. Once the daily quota is exhausted, requests fail with a quota error unless the optional \"Allow paid usage after free quota\" switch is enabled, so this is genuine standing free hosted inference rather than unlimited `$0` usage.",
        "models_detail": "mistralai/Mistral-Small-3.2-24B-Instruct-2506 is the documented recurring free-quota text model, listed as Free on its model page with a context length of 8,192 tokens.",
        "card": "No credit card is required: the Tier 1 page states that registration is free with no credit card required. The optional paid fallback is a separate opt-in setting."
      }
    },
    {
      "id": "free-program-ccb95c320d0c871975c2",
      "program_id": "catalog2-aihubmix",
      "event_type": "program_terms_changed",
      "program_name": "AIHubMix free model catalog",
      "provider_label": "AIHubMix",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://aihubmix.com/models/free",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "55 text models listed at $0 input and $0 output on the provider's free catalog page (updated 2026-08-31); the roster changes over time."
      },
      "after": {
        "models_detail": "56 models listed at $0 input and $0 output on the provider's free catalog page (updated 2026-09-03), nearly all text models plus a few image models (gpt-image-2-free, gemini-3.1-flash-image-preview-free); the roster changes over time."
      }
    },
    {
      "id": "free-program-b0aa8da6134da0e0f3e8",
      "program_id": "catalog2-agnes-ai",
      "event_type": "program_terms_changed",
      "program_name": "Agnes 2.5 Flash at $0",
      "provider_label": "Agnes AI",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.agnes-ai.com/en/docs/pricing",
      "changed_fields": [
        "free_terms",
        "models_detail"
      ],
      "before": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.0-flash and agnes-2.5-flash at $0 per 1M input and output tokens. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to about 20 RPM. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "agnes-2.0-flash and agnes-2.5-flash are currently listed at $0 input and $0 output."
      },
      "after": {
        "free_terms": "Agnes AI's pricing page lists agnes-2.5-flash at $0 per 1M input and output tokens; agnes-2.0-flash is deprecated and kept on the page as a historical record. A new user can create an account and API key without subscribing, then call the hosted OpenAI-compatible API. Free/default text access is rate-limited to about 20 RPM. The pricing page says offers can vary by account eligibility or promotion stage and that the account bill is authoritative.",
        "models_detail": "agnes-2.5-flash is currently listed at $0 input and $0 output; agnes-2.0-flash is marked deprecated, with its $0 prices kept only as historical records."
      }
    },
    {
      "id": "free-program-9f22d232657ead2b7f05",
      "program_id": "gigachat-freemium",
      "event_type": "program_delisted",
      "program_name": "GigaChat Freemium (individuals)",
      "provider_label": "GigaChat (Sber)",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://developers.sber.ru/docs/ru/gigachat/tariffs/individual-tariffs",
      "delisted_status": "uncertain",
      "relabeled": "2026-09-17: was program_ended. The provider did not end the program; we moved the record to uncertain because new clients are routed elsewhere and the tariff page is scoped to existing clients."
    },
    {
      "id": "free-program-7f10884747718ec6f04f",
      "program_id": "ollama-cloud-free-tier",
      "event_type": "program_ended",
      "program_name": "~19 cloud models (deepseek-v4, kimi-k3, gpt-oss-120b...)",
      "provider_label": "Ollama",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://docs.ollama.com/cloud"
    },
    {
      "id": "free-program-7312d4410c13b7b1b14d",
      "program_id": "catalog2-requesty",
      "event_type": "program_terms_changed",
      "program_name": "Requesty free models",
      "provider_label": "Requesty",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.requesty.ai/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Dynamic hosted catalog explicitly identified by Requesty as free models."
      },
      "after": {
        "models_detail": "Requesty's free-models page listed 12 free models on 2026-09-04: NVIDIA Nemotron 3 Ultra, Nemotron 3 Super, Nemotron 3 Nano, Nemotron 3 Nano Omni Reasoning, and Nemotron 3.5 Content Safety; Poolside Laguna M.1 and Laguna XS.2; Google Gemma 4 31B; Mistral Leanstral 1.5; Ling 3.0 Tiny; Muse Glimmer 30B; and one further open model. The docs page carries a shorter nine-row table. Requesty says the set is free for now and announces changes in its changelog."
      }
    },
    {
      "id": "free-program-726f7a8df25263158084",
      "program_id": "ovh-ai-endpoints-free",
      "event_type": "program_terms_changed",
      "program_name": "AI Endpoints (anonymous sandbox)",
      "provider_label": "OVHcloud",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.ovhcloud.com/en/public-cloud/ai-endpoints/",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. The catalog contains over 40 pre-trained models and includes Llama, Qwen, DeepSeek, GPT-OSS, embedding, image-generation, and speech models. Individual models can be retired under OVHcloud's lifecycle process."
      },
      "after": {
        "models_detail": "The anonymous limit is documented offer-wide per model across the current AI Endpoints catalog. The catalog contains over 40 pre-trained models and includes Llama, Qwen, DeepSeek, GPT-OSS, embedding, image-generation, and speech models. Individual models can be retired under OVHcloud's lifecycle process. The authenticated catalog also prices a few entries at Free outright, on 2026-09-04 the beta Qwen3Guard-Gen 0.6B and 8B moderation models (Qwen3.8-27B itself is paid at 0.4 EUR in / 2.7 EUR out per 1M tokens)."
      }
    },
    {
      "id": "free-program-6f0ea13c751c26e84bc3",
      "program_id": "amazon-nova-dev-free",
      "event_type": "program_terms_changed",
      "program_name": "Amazon Nova developer surface",
      "provider_label": "Amazon",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.amazon.com/gp/help/customer/display.html?nodeId=Tsn7s47ytlgjRBHozK",
      "changed_fields": [
        "free_terms",
        "card"
      ],
      "before": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). No numeric limits are published publicly.",
        "card": "not required; no payment terms exist anywhere in the Nova Terms of Use"
      },
      "after": {
        "free_terms": "The Nova Terms of Use (updated 2025-12-02) are a free-use contract with no billing or fee language. Amazon's Nova Act user guide states that API key authentication, using a key from nova.amazon.com/act that is tied to an Amazon.com account, is \"Free of charge with daily limits\"; the alternative AWS IAM authentication path instead applies usage charges to an AWS account. API keys are issued from the developer surface (Terms 1.5), usage may be subject to rate limits or other restrictions at Amazon's discretion (Terms 1.3), and the Build with Nova schedule states the offering does not support deploying production-scale developer applications (S-1.1). No numeric quota or rate-limit figures are published publicly.",
        "card": "not required; no payment terms exist anywhere in the Nova Terms of Use, and the Nova Act user guide describes API key authentication as free of charge with daily limits"
      }
    },
    {
      "id": "free-program-6384a0b06936bc53cd51",
      "program_id": "catalog2-merge-gateway",
      "event_type": "program_terms_changed",
      "program_name": "Merge Gateway free Nemotron route",
      "provider_label": "Merge Gateway",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://docs.merge.dev/merge-gateway/models/details/nvidia-nemotron-3-5-lightning-30b-a3b",
      "changed_fields": [
        "free_terms",
        "card",
        "data_treatment",
        "card_required"
      ],
      "before": {
        "free_terms": "Merge's provider-owned catalog lists `nvidia/nemotron-3.5-lightning-30b-a3b` as available hosted text inference with one NVIDIA host and `Free / Free` input/output pricing. This model-level zero price is separate from Merge's account promotions: the Free plan includes $10 of first-month credits for otherwise paid models, while Pro adds $10 in expiring monthly credits and ordinarily charges model cost plus 5%. Merge also warns that its response `usage.cost` is provider cost rather than necessarily the final invoice, but the exact Nemotron route is explicitly cataloged as free rather than merely lacking pricing metadata.",
        "card": "not_stated",
        "data_treatment": "The reviewed catalog and gateway pages do not establish prompt retention/training treatment for the NVIDIA host; check both Merge and host policies before sensitive use.",
        "card_required": "not_stated"
      },
      "after": {
        "free_terms": "Merge's provider-owned catalog lists `nvidia/nemotron-3.5-lightning-30b-a3b` as available hosted text inference with one NVIDIA host and `Free / Free` input/output pricing. Merge's Free plan (pricing page, 2026-09-04) includes access to all free LLMs, automatic routing with built-in fallback, and starts without a credit card; the earlier $10 first-month credit was removed from the Free tier, while Pro adds $10 in expiring monthly credits and ordinarily charges model cost plus 5%. Merge also warns that its response `usage.cost` is provider cost rather than necessarily the final invoice, but the exact Nemotron route is explicitly cataloged as free rather than merely lacking pricing metadata.",
        "card": "not required; the Free plan says 'Start without a credit card'",
        "data_treatment": "Merge's catalog marks the NVIDIA host for this route 'Zero data retention: No'; neither the catalog nor the gateway pages state whether prompts are used for training, so check both Merge and host policies before sensitive use.",
        "card_required": "no"
      }
    },
    {
      "id": "free-program-617b2b10596b17c419f4",
      "program_id": "google-gemini-free-tier",
      "event_type": "program_terms_changed",
      "program_name": "Gemini API free tier (Flash-class models)",
      "provider_label": "Google",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://ai.google.dev/gemini-api/docs/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Models whose current pricing tables mark standard Free-tier input/output as free of charge include Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash Live Preview, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash-Lite Preview, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding 2, Gemini Embedding, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemini Robotics ER 1.6 Preview, Gemma 4, and two speech-to-text entries (streaming and diarizing transcription) marked free of charge on 2026-09-01. This is model-specific: other listed models and discounted/batch modes may say Free tier 'Not available.'"
      },
      "after": {
        "models_detail": "Models whose current pricing tables mark standard Free-tier input/output as free of charge include Gemini 3.8 Flash (added to the pricing page on 2026-09-03), Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash, Gemini 3.5 Live Translate, Gemini 3.5 Flash-Lite, Gemini 3.1 Flash-Lite, Gemini 3.1 Flash Live Preview, Gemini 3.1 Flash TTS Preview, Gemini 3 Flash Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash-Lite, Gemini 2.5 Flash-Lite Preview, Gemini 2.5 Flash Native Audio, Gemini 2.5 Flash Preview TTS, Gemini 2.0 Flash, Gemini 2.0 Flash-Lite, Gemini Embedding 2, Gemini Embedding, Gemini Robotics ER 2 Preview, Gemini Robotics ER 2 Streaming Preview, Gemini Robotics ER 1.6 Preview, Gemma 4, and two speech-to-text entries (Gemini 3.5 Transcribe Live for streaming and Gemini 3.5 Transcribe for diarizing transcription) marked free of charge on 2026-09-04. This is model-specific: other listed models and discounted/batch modes may say Free tier 'Not available.'"
      }
    },
    {
      "id": "free-program-40d6c29d63469bcaec56",
      "program_id": "siliconflow-cn-free-models",
      "event_type": "program_terms_changed",
      "program_name": "SiliconFlow CN free chat models",
      "provider_label": "SiliconFlow (China)",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://siliconflow.cn/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "Multiple chat models flagged free on the live CN pricing page under Zhipu/GLM, Baidu ERNIE, and Tencent Hunyuan vendor rows; exact IDs are not machine-extractable from the rendered table and should be confirmed on the model marketplace. Free BAAI embeddings and speech models also listed."
      },
      "after": {
        "models_detail": "Free ('\u514d\u8d39') chat models on the CN pricing page on 2026-09-04, several of them behind the page's expand buttons: THUDM/GLM-Z1-9B-0414, THUDM/GLM-4-9B-0414, deepseek-ai/DeepSeek-R1-0528-Qwen3-8B, Qwen/Qwen3.5-4B, Qwen/Qwen3-8B, and Qwen/Qwen2.5-7B-Instruct. Also free: tencent/Hunyuan-MT-7B (translation), PaddlePaddle/PaddleOCR-VL-1.5 and deepseek-ai/DeepSeek-OCR (document OCR), BAAI bge embeddings and rerankers, and several ASR and image models. No Baidu ERNIE chat model is free."
      }
    },
    {
      "id": "free-program-23e8591f16a2510c17a8",
      "program_id": "mancer-free-models",
      "event_type": "program_terms_changed",
      "program_name": "Mancer free test models",
      "provider_label": "Mancer",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://mancer.tech/pricing",
      "changed_fields": [
        "models_detail"
      ],
      "before": {
        "models_detail": "MythoLite only: a LLaMA 2 13B creative/roleplay model with a 2,560-token context limit and a 150-token maximum output on the current model table."
      },
      "after": {
        "models_detail": "MythoLite only: on 2026-09-04 the Mancer model table listed MythoLite as a LLaMA 2 13B model tagged Creative and Roleplay, with both input and output token cost shown as FREE. The model table and model detail panel do not currently display a context length or maximum output figure for MythoLite."
      }
    },
    {
      "id": "free-program-0aabb4cb58b9930ea818",
      "program_id": "opencode-zen-free",
      "event_type": "program_terms_changed",
      "program_name": "Seven free models on the Zen gateway",
      "provider_label": "OpenCode",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://opencode.ai/docs/zen/",
      "changed_fields": [
        "free_terms",
        "models_detail",
        "data_treatment"
      ],
      "before": {
        "free_terms": "OpenCode Zen currently prices six models at Free for input, output, and cached-read tokens. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the current Zen pricing table: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, and Muse Spark 1.2 Contributor Free. Ox Alpha Free and Hy3 Free were removed from the free table before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.2 Contributor Free permits prompts and completions to train future Meta models."
      },
      "after": {
        "free_terms": "OpenCode Zen currently prices seven models at Free for input, output, and cached-read tokens. The provider describes every free model as available for a limited time and publishes no request quota or end date. Zen's setup flow says to sign in, add billing details, and copy an API key.",
        "models_detail": "Free rows on the current Zen pricing table: Big Pickle, MiMo-V2.5 Free, Ling 3.0 Flash Fin Free, Nemotron 3 Ultra Free, Nemotron 3.5 Lightning Free, Muse Spark 1.3 Contributor Free (added 2026-09-04), and Muse Spark 1.2 Contributor Free. Ox Alpha Free and Hy3 Free were removed from the free table before 2026-09-01. Other Zen models are pay-as-you-go.",
        "data_treatment": "Model-specific. Big Pickle, MiMo-V2.5 Free, and Ling 3.0 Flash Fin Free may use collected data to improve their models during the free period. Nemotron 3 Ultra Free and Nemotron 3.5 Lightning Free log use for security and to improve NVIDIA products and services. Muse Spark 1.3 Contributor Free and Muse Spark 1.2 Contributor Free permit prompts and completions to train future Meta models."
      }
    },
    {
      "id": "free-program-06bf18f2ff4d25ffc4bc",
      "program_id": "upstage-ai-support-program",
      "event_type": "program_terms_changed",
      "program_name": "Solar Pro free for eligible institutions",
      "provider_label": "Upstage",
      "detected_at": "2026-09-04T00:00:00+00:00",
      "date": "2026-09-04",
      "evidence_url": "https://www.upstage.ai/pricing/api",
      "changed_fields": [
        "clock"
      ],
      "before": {
        "clock": "Applications are accepted through 2027-03-31 23:00 KST; approved access lasts up to one year from approval."
      },
      "after": {
        "clock": "The program page lists the event period as running through March 31, 2027 11:00 PM (KST). Approved organizations have free access from the date of approval until March 31, 2027, for up to one year, so that date caps access rather than only the application window."
      }
    },
    {
      "id": "free-event-d9a76cb8615c40a9403c",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "inclusionai/ling-3.0-flash-fin:free",
      "model": "Ling 3.0 Flash Fin",
      "model_key": "ling30flashfinfree",
      "detected_at": "2026-08-28T11:30:09+00:00",
      "date": "2026-08-28",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-fin:free"
    },
    {
      "id": "free-event-c56a98e8b79255cb9cde",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nvidia/nemotron-nano-12b-v2-vl:free",
      "model": "NVIDIA: Nemotron Nano 12B 2 VL",
      "model_key": "nemotronnano12bv2vlfree",
      "detected_at": "2026-08-25T11:30:09+00:00",
      "date": "2026-08-25",
      "evidence_url": "https://openrouter.ai/nvidia/nemotron-nano-12b-v2-vl:free"
    },
    {
      "id": "free-event-83a6e8a1d78df1a362ef",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nvidia/nemotron-3-nano-30b-a3b:free",
      "model": "NVIDIA: Nemotron 3 Nano 30B A3B",
      "model_key": "nemotron3nano30ba3bfree",
      "detected_at": "2026-08-25T11:30:09+00:00",
      "date": "2026-08-25",
      "evidence_url": "https://openrouter.ai/nvidia/nemotron-3-nano-30b-a3b:free"
    },
    {
      "id": "free-event-5447ee242da1b6eb0d65",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "minimax/minimax-m2.7:free",
      "model": "MiniMax: MiniMax M2.7",
      "model_key": "minimaxm27free",
      "detected_at": "2026-08-25T11:30:09+00:00",
      "date": "2026-08-25",
      "evidence_url": "https://openrouter.ai/minimax/minimax-m2.7:free"
    },
    {
      "id": "free-event-427a7d14c6daee6986e7",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nvidia/nemotron-nano-9b-v2:free",
      "model": "NVIDIA: Nemotron Nano 9B V2",
      "model_key": "nemotronnano9bv2free",
      "detected_at": "2026-08-25T11:30:09+00:00",
      "date": "2026-08-25",
      "evidence_url": "https://openrouter.ai/nvidia/nemotron-nano-9b-v2:free"
    },
    {
      "id": "free-event-4079458c2533ed4193f5",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "minimax/minimax-m3:free",
      "model": "MiniMax: MiniMax M3",
      "model_key": "minimaxm3free",
      "detected_at": "2026-08-25T11:30:09+00:00",
      "date": "2026-08-25",
      "evidence_url": "https://openrouter.ai/minimax/minimax-m3:free"
    },
    {
      "id": "free-event-e3dbf0e959b9ffcea873",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "openai/gpt-oss-20b:free",
      "model": "OpenAI: gpt-oss-20b",
      "model_key": "gptoss20bfree",
      "detected_at": "2026-08-22T11:30:10+00:00",
      "date": "2026-08-22",
      "evidence_url": "https://openrouter.ai/openai/gpt-oss-20b:free"
    },
    {
      "id": "free-event-a11976ad89ef064623fc",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "thinkingmachines/inkling:free",
      "model": "Thinking Machines: Inkling",
      "model_key": "inklingfree",
      "detected_at": "2026-08-22T11:30:10+00:00",
      "date": "2026-08-22",
      "evidence_url": "https://openrouter.ai/thinkingmachines/inkling:free"
    },
    {
      "id": "free-event-0d32fa7d1170f3fcd571",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "thinkingmachines/inkling-small:free",
      "model": "Thinking Machines: Inkling Small",
      "model_key": "inklingsmallfree",
      "detected_at": "2026-08-22T11:30:10+00:00",
      "date": "2026-08-22",
      "evidence_url": "https://openrouter.ai/thinkingmachines/inkling-small:free"
    },
    {
      "id": "free-event-e88bf196fa74845a1a83",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "z-ai/glm-5.2:free",
      "model": "Z.ai: GLM 5.2",
      "model_key": "glm52free",
      "detected_at": "2026-08-19T11:30:15+00:00",
      "date": "2026-08-19",
      "evidence_url": "https://openrouter.ai/z-ai/glm-5.2:free"
    },
    {
      "id": "free-event-e80dcf3aea67527ba160",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "liquid/lfm-2.5-2.6b:free",
      "model": "LiquidAI: LFM2.5-2.6B",
      "model_key": "lfm2526bfree",
      "detected_at": "2026-08-19T11:30:15+00:00",
      "date": "2026-08-19",
      "evidence_url": "https://openrouter.ai/liquid/lfm-2.5-2.6b:free"
    },
    {
      "id": "free-event-750c4c0dc3e7b968914a",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "dots-studio/dots-3-note-preview:free",
      "model": "Dots Studio: Dots3-Note Preview",
      "model_key": "dots3notepreviewfree",
      "detected_at": "2026-08-15T11:30:09+00:00",
      "date": "2026-08-15",
      "evidence_url": "https://openrouter.ai/dots-studio/dots-3-note-preview:free"
    },
    {
      "id": "free-event-bd5a93893b6e1e5d732c",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "inclusionai/ling-3.0-tiny:free",
      "model": "inclusionAI: Ling 3.0 Tiny",
      "model_key": "ling30tinyfree",
      "detected_at": "2026-08-13T11:30:09+00:00",
      "date": "2026-08-13",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-tiny:free"
    },
    {
      "id": "free-event-ba8f18053a6b0cd31ce8",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "nvidia/nemotron-3.5-lightning:free",
      "model": "NVIDIA: Nemotron 3.5 Lightning",
      "model_key": "nemotron35lightningfree",
      "detected_at": "2026-08-13T11:30:09+00:00",
      "date": "2026-08-13",
      "evidence_url": "https://openrouter.ai/nvidia/nemotron-3.5-lightning:free"
    },
    {
      "id": "free-event-1939ffef6a7b8b4f404c",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "liquid/lfm-2.5-2.6b:free",
      "model": "LiquidAI: LFM2.5-2.6B",
      "model_key": "lfm2526bfree",
      "detected_at": "2026-08-13T11:30:09+00:00",
      "date": "2026-08-13",
      "evidence_url": "https://openrouter.ai/liquid/lfm-2.5-2.6b:free"
    },
    {
      "id": "free-event-e205835ebc511549446a",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "inclusionai/ling-3.0-flash:free",
      "model": "Ling-3.0-flash",
      "model_key": "ling30flashfree",
      "detected_at": "2026-08-07T11:30:16+00:00",
      "date": "2026-08-07",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash:free"
    },
    {
      "id": "free-event-5603c45a63ece2922c83",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "inclusionai/ling-3.0-tiny:free",
      "model": "inclusionAI: Ling 3.0 Tiny",
      "model_key": "ling30tinyfree",
      "detected_at": "2026-08-07T11:30:16+00:00",
      "date": "2026-08-07",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-tiny:free"
    },
    {
      "id": "free-event-8c5b2473b81a2be7e3b1",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "poolside/laguna-m.1:free",
      "model": "Poolside: Laguna M.1",
      "model_key": "lagunam1free",
      "detected_at": "2026-07-29T11:30:13+00:00",
      "date": "2026-07-29",
      "evidence_url": "https://openrouter.ai/poolside/laguna-m.1:free"
    },
    {
      "id": "free-event-c65612b19df09603478a",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "inclusionai/ling-3.0-flash:free",
      "model": "Ling-3.0-flash",
      "model_key": "ling30flashfree",
      "detected_at": "2026-07-24T11:30:13+00:00",
      "date": "2026-07-24",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash:free"
    },
    {
      "id": "free-event-711c8a2937f9d8b10ca8",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "poolside/laguna-s-2.1:free",
      "model": "Poolside: Laguna S 2.1",
      "model_key": "lagunas21free",
      "detected_at": "2026-07-22T11:30:12+00:00",
      "date": "2026-07-22",
      "evidence_url": "https://openrouter.ai/poolside/laguna-s-2.1:free"
    },
    {
      "id": "free-event-1da0477adcb58b78226d",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "tencent/hy3:free",
      "model": "Tencent: Hy3",
      "model_key": "hy3free",
      "detected_at": "2026-07-21T11:30:12+00:00",
      "date": "2026-07-21",
      "evidence_url": "https://openrouter.ai/tencent/hy3:free"
    },
    {
      "id": "free-event-bc57f4b4f03cf57b31d9",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "meta-llama/llama-3.2-3b-instruct:free",
      "model": "Meta: Llama 3.2 3B Instruct",
      "model_key": "llama323binstructfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/meta-llama/llama-3.2-3b-instruct:free"
    },
    {
      "id": "free-event-7f44168f9eda77ad21de",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "meta-llama/llama-3.3-70b-instruct:free",
      "model": "Meta: Llama 3.3 70B Instruct",
      "model_key": "llama3370binstructfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/meta-llama/llama-3.3-70b-instruct:free"
    },
    {
      "id": "free-event-2704064d4f9d74f47eb7",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "qwen/qwen3-next-80b-a3b-instruct:free",
      "model": "Qwen: Qwen3 Next 80B A3B Instruct",
      "model_key": "qwen3next80ba3binstructfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/qwen/qwen3-next-80b-a3b-instruct:free"
    },
    {
      "id": "free-event-2439f7883221c7de7aa3",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nousresearch/hermes-3-llama-3.1-405b:free",
      "model": "Nous: Hermes 3 405B Instruct",
      "model_key": "hermes3llama31405bfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/nousresearch/hermes-3-llama-3.1-405b:free"
    },
    {
      "id": "free-event-216a72435618d0b8ff19",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "qwen/qwen3-coder:free",
      "model": "Qwen: Qwen3 Coder 480B A35B",
      "model_key": "qwen3coderfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/qwen/qwen3-coder:free"
    },
    {
      "id": "free-event-0b95f3ea6529c4771f4d",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "cognitivecomputations/dolphin-mistral-24b-venice-edition:free",
      "model": "Venice: Uncensored",
      "model_key": "dolphinmistral24bveniceeditionfree",
      "detected_at": "2026-07-20T11:30:12+00:00",
      "date": "2026-07-20",
      "evidence_url": "https://openrouter.ai/cognitivecomputations/dolphin-mistral-24b-venice-edition:free"
    },
    {
      "id": "free-event-bc1e138745eb2d851d6b",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "liquid/lfm-2.5-1.2b-thinking:free",
      "model": "LiquidAI: LFM2.5-1.2B-Thinking",
      "model_key": "lfm2512bthinkingfree",
      "detected_at": "2026-07-14T11:30:08+00:00",
      "date": "2026-07-14",
      "evidence_url": "https://openrouter.ai/liquid/lfm-2.5-1.2b-thinking:free"
    },
    {
      "id": "free-event-b11caae705e420c0e125",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "liquid/lfm-2.5-1.2b-instruct:free",
      "model": "LiquidAI: LFM2.5-1.2B-Instruct",
      "model_key": "lfm2512binstructfree",
      "detected_at": "2026-07-14T11:30:08+00:00",
      "date": "2026-07-14",
      "evidence_url": "https://openrouter.ai/liquid/lfm-2.5-1.2b-instruct:free"
    },
    {
      "id": "free-event-55661d35478713764486",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "openai/gpt-oss-120b:free",
      "model": "OpenAI: gpt-oss-120b",
      "model_key": "gptoss120bfree",
      "detected_at": "2026-07-14T11:30:08+00:00",
      "date": "2026-07-14",
      "evidence_url": "https://openrouter.ai/openai/gpt-oss-120b:free"
    },
    {
      "id": "free-event-22e2ef9695e57490ed9d",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "poolside/laguna-xs.2:free",
      "model": "Poolside: Laguna XS.2",
      "model_key": "lagunaxs2free",
      "detected_at": "2026-07-10T11:30:09+00:00",
      "date": "2026-07-10",
      "evidence_url": "https://openrouter.ai/poolside/laguna-xs.2:free"
    },
    {
      "id": "free-event-8e12ffe5e04dd428dfe0",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "tencent/hy3:free",
      "model": "Tencent: Hy3",
      "model_key": "hy3free",
      "detected_at": "2026-07-07T11:30:14+00:00",
      "date": "2026-07-07",
      "evidence_url": "https://openrouter.ai/tencent/hy3:free"
    },
    {
      "id": "free-event-418f3dfb1e364666661c",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "poolside/laguna-xs-2.1:free",
      "model": "Poolside: Laguna XS 2.1",
      "model_key": "lagunaxs21free",
      "detected_at": "2026-07-03T11:30:10+00:00",
      "date": "2026-07-03",
      "evidence_url": "https://openrouter.ai/poolside/laguna-xs-2.1:free"
    },
    {
      "id": "free-event-9a20be72228cadf28bd6",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "cohere/north-mini-code:free",
      "model": "Cohere: North Mini Code",
      "model_key": "northminicodefree",
      "detected_at": "2026-07-02T11:30:09+00:00",
      "date": "2026-07-02",
      "evidence_url": "https://openrouter.ai/cohere/north-mini-code:free"
    },
    {
      "id": "free-event-5d42a33f26f2f443fd2f",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "nex-agi/nex-n2-pro:free",
      "model": "Nex AGI: Nex-N2-Pro",
      "model_key": "nexn2profree",
      "detected_at": "2026-07-02T11:30:09+00:00",
      "date": "2026-07-02",
      "evidence_url": "https://openrouter.ai/nex-agi/nex-n2-pro:free"
    }
  ],
  "pending": [
    {
      "id": "free-event-e6bd02d4a42029801520",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "inclusionai/ling-3.0-flash-sante:free",
      "model": "inclusionAI: Ling 3.0 Flash Sante",
      "model_key": "ling30flashsantefree",
      "detected_at": "2026-10-09T11:30:13+00:00",
      "date": "2026-10-09",
      "evidence_url": "https://openrouter.ai/inclusionai/ling-3.0-flash-sante:free"
    }
  ],
  "suppressed": [
    {
      "id": "free-event-67f13698f5147c072625",
      "program_id": "openrouter-free-program",
      "event_type": "model_added",
      "model_id": "z-ai/glm-5.2:free",
      "model": "Z.ai: GLM 5.2",
      "model_key": "glm52free",
      "detected_at": "2026-08-17T11:30:09+00:00",
      "date": "2026-08-17",
      "evidence_url": "https://openrouter.ai/z-ai/glm-5.2:free"
    },
    {
      "id": "free-event-339d78f0b6d81c341366",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "liquid/lfm-2.5-2.6b:free",
      "model": "LiquidAI: LFM2.5-2.6B",
      "model_key": "lfm2526bfree",
      "detected_at": "2026-08-18T11:30:11+00:00",
      "date": "2026-08-18",
      "evidence_url": "https://openrouter.ai/liquid/lfm-2.5-2.6b:free"
    },
    {
      "id": "free-event-a66d8a6ed9a79e58fc8f",
      "program_id": "openrouter-free-program",
      "event_type": "model_removed",
      "model_id": "z-ai/glm-5.2:free",
      "model": "Z.ai: GLM 5.2",
      "model_key": "glm52free",
      "detected_at": "2026-08-18T11:30:11+00:00",
      "date": "2026-08-18",
      "evidence_url": "https://openrouter.ai/z-ai/glm-5.2:free"
    }
  ]
}