freellmpool

Our take

Python CLI, library, and local OpenAI-compatible proxy that pools the free tiers of 19 LLM providers (Groq, Cerebras, NVIDIA NIM, Gemini, OpenRouter, GitHub Models, and others) with automatic failover and persistent per-day quota tracking; freellmpool mcp exposes ask, panel/second-opinion, tokenmax, route-preview, models, quota, and stats tools over stdio, and it can start keyless via Pollinations, OVHcloud, or Kilo Gateway. Thoughtfully engineered (CI, Tailscale gateway mode, install profiles for Claude Code, aider, Cursor), MIT-licensed, but the repo is only a month old with essentially one maintainer, and squeezing pooled free tiers sits in a ToS gray area its own FAQ acknowledges — provider bans are a real possibility, and the MCP server is one feature of a larger tool, not a standalone server. For hobbyists and cost-conscious tinkerers running experimental agent workloads; not something to build production systems on.

reviewed by hand · 2026-07-05

Something wrong or dead here? Report it

Verification record

last verified

12 days ago

github stars

60

last commit

13 days ago

archived

no

license

MIT

in registry since

2026-06-17

github.com/0xzr/freellmpool