freellmapi is a self-hosted, OpenAI-compatible local /v1 endpoint that aggregates free tiers from 34 LLM providers behind one API with smart routing, automatic failover, and encrypted provider keys. It runs locally without a GPU and includes a web dashboard for key management and analytics.
Project overview
Aggregating roughly 7.4B free tokens per month across 34 providers behind a single OpenAI-compatible endpoint, with automatic failover on rate limits, per-key AES-256-GCM encryption in SQLite, and a self-updating signed model catalog, addresses the recurring pain of manually stacking ever-changing free-tier provider quotas.
Project type
Model Runtime
Use cases
Coding & Development
Deployment
Refer to project documentation
License
MIT
Best for
Developers doing personal experimentation who want to use free-tier LLM quotas from many providers through one local, OpenAI-compatible endpoint.
Key capabilities
Live per-model speed/capability/reliability scores rank the fallback chain; automatic failover retries the next model on 429/5xx errors with cooldowns and key rotation, including per-key rate tracking.
A React UI to manage encrypted provider keys, reorder the fallback chain, run a playground, and read p50/p95/TTFT analytics.
Limitations and risks
No frontier models, variable latency, and no SLA; intelligence dips late in the day as top models hit their daily caps.
The repository description states the project is intended for personal experimentation only; consider this before adopting it for broader use.
Getting started
Setup is rated easy and requires no coding. The project offers a web GUI for management (including adding provider API keys on the Keys page) and an OpenAI-compatible API for clients.
Evidence and sources
GitHub project description: 7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic fa…
README: The problem is that stacking them by hand is painful: thirty-four different SDKs, thirty-four different rate limits, thirty-four places a request can fail.
README: FreeLLMAPI collapses that into one OpenAI-compatible endpoint. Point any OpenAI client library at your local server, and it routes transparently across whichever providers you've…
README: live per-model speed/capability/reliability scores rank your chain; automatic fallover retries the next model on 429/5xx with cooldowns and key rotation