160
One endpoint for Groq, Gemini, Mistral, Cerebras, and Ollama. ~80 free requests per minute combined. If one provider is rate-limited or down, your request routes to the next one automatically.
Three meta-models: free-fast (speed), free-smart (capability), free (availability). Drop-in replacement for the OpenAI SDK. Change your base URL, keep your code.
Built-in circuit breakers, sliding-window rate tracking, per-client limits, Zod validation, and a real-time React dashboard. Docker one-command deploy. MIT licensed.
Built with