OVHcloud AI Endpoints
OVHcloud AI Endpoints offers a permanent free anonymous tier — no signup or API key — to call 40+ EU-hosted open-weight models (Llama, Qwen, Mixtral, DeepSeek-R1 distill) via an OpenAI-compatible API.
- Requests / min
- 2
- Card required
- No
OVHcloud AI Endpoints scores 6.2/10 for free value in LLM. The main constraint: Free anonymous tier of 2 requests/minute per IP per model, no API key or signup needed (grab a free token on the site). Higher limits require a Public Cloud account and are pay-as-you-go.
Free value score
6.2 / 10Show the math
Key metrics
Models
24 free · 24 totalLimits may be stale — last re-verified outside this section's freshness window.
| Model | Family | AA | Access |
|---|---|---|---|
| Qwen3.6-27B | Qwen | 37 | free |
| Qwen3.5-397B-A17B | Qwen | 32 | free |
| gpt-oss-120b | GPT | 23.8 | free |
| gpt-oss-20b | GPT | 14.9 | free |
| bge-m3 | bge | — | free |
| bge-multilingual-gemma2 | Gemma | — | free |
| Meta-Llama-3_3-70B-Instruct | Llama | — | free |
| Mistral-7B-Instruct-v0.3 | Mistral | — | free |
| Mistral-Nemo-Instruct-2407 | Mistral | — | free |
| Mistral-Small-3.2-24B-Instruct-2506 | Mistral | — | free |
| nvr-tts-de-de | NVIDIA Riva | — | free |
| nvr-tts-en-us | NVIDIA Riva | — | free |
| nvr-tts-es-es | NVIDIA Riva | — | free |
| nvr-tts-it-it | NVIDIA Riva | — | free |
| Qwen2.5-VL-72B-Instruct | Qwen | — | free |
| Qwen3-32B | Qwen | — | free |
| Qwen3-Coder-30B-A3B-Instruct | Qwen | — | free |
| Qwen3-Embedding-8B | Qwen | — | free |
| Qwen3.5-9B | Qwen | — | free |
| Qwen3Guard-Gen-0.6B | Qwen | — | free |
| Qwen3Guard-Gen-8B | Qwen | — | free |
| stable-diffusion-xl-base-v10 | Stable Diffusion | — | free |
| whisper-large-v3 | Whisper | — | free |
| whisper-large-v3-turbo | Whisper | — | free |
What you get
Truly no-signup free tier — anonymous access with a free token
40+ open-weight models hosted in EU data centers
OpenAI-SDK-compatible endpoint
First-party offering from a major European cloud provider
Gotchas
No gotchas documented yet. Contribute one if you know a constraint we missed.
Sources
-
"Anonymous: 2 requests per minute, per IP and per model. Authenticated API-key access has a higher project-level limit."
https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-getting-started -
"As of now, the AI Endpoints platform does not impose any usage limits for API requests, apart from the rate and payload size limiting."
https://docs.ovhcloud.com/en/guides/public-cloud/ai-machine-learning/ai-endpoints-capabilities -
"Free tier (with rate limits): No API key required. You can omit the apiKey parameter or set it to an empty string"
https://developers.llamaindex.ai/typescript/framework/modules/models/llms/ovhcloud/ -
"gpt-oss-120b (high) achieves a score of 33 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding."
https://artificialanalysis.ai/models/gpt-oss-120b
Alternatives
Same category · similar quota
LEGACY launch-blog tier: 125 requests/5h, all open models, unlimited tokens, UI + API. Current public pricing sells $30 packs (500 req/5h + $24 credits/week) — see Synthetic Pack
Two stacked packs: 1,000 requests/5h + $48 API credits/week + 2 concurrent requests per model across GLM-5.2 (512k), Kimi-K2.7-Code, MiniMax M3 and other open models; native Anthropic + OpenAI endpoints