Free LLM APIs

All free tiers / OVHcloud

OVHcloud AI Endpoints free tier

Every AI Endpoints model can be called for free without an API key, at a low anonymous rate limit. OpenAI-compatible API at https://oai.endpoints.kepler.ai.cloud.ovh.net/v1.

Kind: Free tier with rate limits · Provider type: Inference host · EU provider

Card needed: no (stated by the provider)

Limits: Anonymous (no key): 2 requests per minute, per IP and per model. With an API access key (paid, billed per token): 400 requests per minute per Public Cloud project and model. Payload limit 2 MB (10 MB for vision models).

Your data: the free-tier page doesn't say

Free models and offers (22)

ModelRequestsTokensOther limits / notes
Meta-Llama-3_3-70B-Instruct2/minn/panonymous, per IP
Mistral-Nemo-Instruct-24072/minn/panonymous, per IP
Mistral-Small-3.2-24B-Instruct-25062/minn/panonymous, per IP
Qwen2.5-VL-72B-Instruct2/minn/panonymous, per IP
Qwen3-Embedding-8B2/minn/panonymous, per IP
Qwen3.5-397B-A17B2/minn/panonymous, per IP
Qwen3.5-9B2/minn/panonymous, per IP
Qwen3.6-27B2/minn/panonymous, per IP
Qwen3.8-27B2/minn/panonymous, per IP
Qwen3Guard-Gen-0.6B2/minn/panonymous, per IP
Qwen3Guard-Gen-8B2/minn/panonymous, per IP
bge-m32/minn/panonymous, per IP
bge-multilingual-gemma22/minn/panonymous, per IP
gpt-oss-120b2/minn/panonymous, per IP
gpt-oss-20b2/minn/panonymous, per IP
nvr-tts-de-de2/minn/panonymous, per IP
nvr-tts-en-us2/minn/panonymous, per IP
nvr-tts-es-es2/minn/panonymous, per IP
nvr-tts-it-it2/minn/panonymous, per IP
stable-diffusion-xl-base-v102/minn/panonymous, per IP
whisper-large-v32/minn/panonymous, per IP
whisper-large-v3-turbo2/minn/panonymous, per IP

n/p = not published on the provider's page. Used for training: not stated.

Notes

Sources

Need more than the free tier? Compare paid per-token prices on prices.branolabs.com.