The free LLM API model catalog
Every model here costs $0 through FreeLLMAPI, a free LLM API with one OpenAI-compatible key that routes to whichever provider serves each model free.
395 models across 38 providers, every one of them $0 through FreeLLMAPI, a free LLM API with one OpenAI-compatible key that sends each request to whichever provider serves the model free. The deepest catalogs right now are ElectronHub (65), Cloudflare Workers AI (51), AI Horde (32), xKiro (28) and Cohere (24).
Each provider meters its own free tier, usually by requests per minute and requests per day, sometimes by tokens per minute or per day, and every model page below lists those exact limits, the context window, and the known quirks for each provider that serves it. The router counts that usage per key and fails over to the next provider when one is rate-limited or out of daily quota.
335 of 335 models · 438 free endpoints
Every free model, A to Z
The whole catalog as a plain list — every model family we track, each linking to its providers, free limits and quirks.
Free LLM API questions
What is a free LLM API?
An endpoint that serves real model inference on a provider's permanent no-payment tier, metered by requests or tokens per minute and per day rather than by a credit balance. FreeLLMAPI is one: an open-source router you run on your own machine that puts every free tier in this catalog behind a single OpenAI-compatible endpoint. Routing costs nothing; the only thing sold is the optional Premium catalog feed.
How does FreeLLMAPI keep the API free?
It only calls each provider's own free tier, using keys you add yourself, so inference costs nothing and your keys never leave your machine. Premium ($19 a year or $49 once) keeps the model catalog live on your router; free installs pull the monthly snapshot, in which a model appears 30 days after it joins the live feed.
Do I need a credit card?
Not for FreeLLMAPI. Each provider issues its own free key, and most tiers in this catalog issue one without a card; each model page lists the limits and sign-up quirks of every provider that serves it.
What happens when a provider drops a free model?
Your router pulls a signed catalog twice a day, and a model we have removed or disabled leaves your router at the next sync. Between syncs, a provider that answers a request with an end-of-life error gets that model retired automatically, and the request fails over to the next model in your chain. Details: what happens when your free models get removed.
Is it OpenAI-compatible?
Yes. Point any OpenAI SDK at http://localhost:3001/v1 with your unified key. The router also speaks the Anthropic Messages surface and Gemini's native wire on /v1beta, and every response carries an X-Routed-Via header naming the provider and model that served it.
This catalog is live. Is yours?
Premium routers pull the live feed: new free models, quota changes, and compatibility fixes arrive as soon as we ship them.
Go live · $19/yr →