FreeLLM API
A server you install on your machine that exposes one OpenAI and Anthropic compatible address in front of the free tiers of dozens of model providers. You drop in one key per provider, it encrypts them, and it routes each request to the best-scored model that still has quota; when a provider answers 'limit reached' or fails, it moves to the next. The repository claims 10.1 billion tokens a month, 87 providers and 1,313 free endpoints as of 16 November 2026: that is the sum of declared free quotas, not what a single account will get. One command plugs Claude Code into it, and that is the trap: Claude Code then runs on models other than Claude, picked by the router. The project describes itself as meant for personal experimentation, not production.