Cut Your LLM Bill to Zero with a Free Multi-Provider Proxy
Every AI project hits the same wall: free-tier APIs are great but unreliable, and paid APIs eat your runway.
The problem
Free tiers from Groq, Mistral, Gemini and others are genuinely useful — but each one has limits, occasional outages, and different rate caps. Chaining them by hand is tedious.
The solution: DAVIL Cod
DAVIL Cod is an open-source (MIT) LLM proxy that rotates 13+ free providers behind one OpenAI-compatible endpoint. If one dies, your request falls through to the next.
npx davil-cod init
npx davil-cod start
Key features
- Circuit breaker — a dead provider never kills your app
- Dashboard — provider status, RPM graphs, remaining limits
- Disk cache — identical prompts don't burn limits twice
- Docker — single container
Works with anything
Point any OpenAI-compatible client at http://localhost:4000/v1 — opencode, Cursor, your own scripts.
FAQ
Is it really free? Yes — it routes to free model tiers. You only need API keys for those free tiers.
Does it support my provider? Groq, Mistral, Gemini, NVIDIA NIM, OpenRouter, ZAI and more — extensible via providers.json.
