Switch models without losing your context

Hey! 👋 Here's what I promised you:

Every time you hit your Claude limits and switch to another model, you end up explaining your project again from zero. FreeLLMAPI fixes that.

FreeLLMAPI — a free, open-source router (MIT) that connects Claude Code to many free models — Gemini, GLM, Mistral, Qwen and more — behind one setup. When one model hits its limit, it moves on to the next one by itself, so your session keeps going.

The difference: the handoff note. Before switching, it adds a short note for the new model — "you are taking over an ongoing task, don't restart it" plus a summary of the recent chat — so it understands where you stopped without you re-explaining anything. ⚠️ It's off by default: turn it on in your .env:

FREELLMAPI_CONTEXT_HANDOFF=on_model_switch

Setup

1. Install it. On Windows, the easiest way is the .exe installer from Releases:

On macOS/Linux (Docker required):

curl -fsSL https://freellmapi.co/install.sh | bash

2. Open the dashboard and add the free API keys of the providers you want to use (Google, Mistral, OpenRouter, NVIDIA…).
3. Connect Claude Code with the unified key from the dashboard:

npx freellmapi setup-claude --url http://localhost:3001 --api-key <unified-key>

Be honest with yourself

• "Free" depends on third-party free tiers: each provider sets its own limits and can change or end them at any time.
• The project itself says it's for personal experimentation and learning, not production — no frontier models, variable speed, no SLA.
• It stores your provider API keys (encrypted) — read the repo before you trust it, and the terms you accepted with each provider still apply.
• The router is free; the paid "Premium" plan only gets you new models in the catalog faster.