A snapshot of the free LLM API landscape, drawn from a continuously-maintained catalog of 27+ providers and 310+ free model endpoints.
Key findings
310+ free model endpoints are available right now across 27+ providers, without payment.
Most free tiers are rate-limited, not credit-based. The majority cap by requests/day or tokens/day rather than a dollar balance.
True recurring monthly credits are rare. Reka (~$10/mo) is one of the few; most "$X free" offers are one-time signup bonuses.
Some "free" APIs are gone or need a card. SambaNova has retired its free tier entirely; Vercel AI Gateway requires a card.
Open weights dominate. The same open models appear across many providers, so availability is resilient to any one provider changing terms.
Open-model ubiquity
The most-available model families, by number of providers serving them free:
Model family
Free availability
Llama
12 providers
Gemma
10 providers
Qwen
8 providers
GPT-OSS
8 providers
Nemotron
8 providers
GLM
7 providers
Mistral
6 providers
DeepSeek
5 providers
Methodology
Figures are taken from the FreeLLMAPI live catalog, which tracks each provider's publicly documented free tier and is verified against live API responses. The landscape shifts weekly as providers launch, retire, or re-price models — see the live catalog for the current state.
Browse every free model by name
The full set of free model families tracked right now — each links to its providers, limits, and quirks.
Use every one of these through one key. FreeLLMAPI is an open-source, self-hosted router that puts all these free tiers behind a single OpenAI-compatible endpoint and fails over when one is rate-limited. Browse the catalog or go live.