OVHcloud: AI Endpoints (anonymous sandbox)

The terms in full

Anonymous AI Endpoints access is available without an API key at 2 requests per minute per IP per model. Authenticated use is a separate billed path with a 400-requests-per-minute per-project per-model limit and requires an OVHcloud Public Cloud project with a payment method. Playground LLM responses are limited to 1,024 output tokens for testing.

• When it runs out Requests are rejected (429) until the window resets.
• Clock No fixed end date for anonymous free access is stated. Individual models have a lifecycle: announced retirements receive a three-month grace period before removal.
• Region No geographic restriction on anonymous use is stated. OVHcloud's worldwide product page lists the Base and Batch APIs with "Worldwide availability (non-EU)"; authenticated project and catalog availability may vary by OVHcloud region.

Models

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.

Signup, payment, data

• Signup not required for anonymous access; required for authenticated API-key access
• Card / payment not required for anonymous access; a payment method is required for authenticated Public Cloud access, and Discovery-mode projects without one cannot use the service
• Training Provider says it does not train on your prompts or outputs.
• Data terms 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.

What the provider's page says

Every structured number above is backed by one of these quotes, checked verbatim against the page our crawler fetched. A field with no quote is shown as not stated.

• payment “Anonymous: 2 requests per minute, per IP and per model.” source
• limits “Anonymous: 2 requests per minute, per IP and per model.” source
• at exhaustion “If you exceed this limit, a 429 error code will be returned.” source

Evidence pages, re-read nightly

• https://www.ovhcloud.com/en/public-cloud/ai-endpoints/
• https://endpoints.ai.cloud.ovh.net/
• https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-getting-started
• https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-billing
• https://www.ovhcloud.com/en-gb/public-cloud/ai-endpoints/catalog/
• https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-capabilities

Change history

1. 2026-10-09 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail, data treatment)
   • 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 reti
   • 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.
2. 2026-10-08 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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 fami
3. 2026-10-07 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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 gener
4. 2026-10-05 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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, imag
5. 2026-10-02 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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
6. 2026-10-01 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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,
7. 2026-09-30 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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. Individu
8. 2026-09-28 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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.
9. 2026-09-25 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
   • 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-di
10. 2026-09-24 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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 speec
11. 2026-09-23 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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 Llam
12. 2026-09-22 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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 embeddin
13. 2026-09-21 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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 em
14. 2026-09-17 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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, Qwen3Guar
15. 2026-09-04 Terms changed: OVHcloud AI Endpoints (anonymous sandbox) (models detail)
    • 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 pro