Analysis updated 2026-06-21
Use Claude Code's interface with a free OpenRouter model instead of paying Anthropic.
Run Claude Code pointed at a local Llama or DeepSeek model via Ollama for full privacy.
Set up a Discord or Telegram bot that runs AI coding sessions using any free AI backend.
Route different Claude model tiers to different providers based on cost or capability.
| alishahryar1/free-claude-code | mkdocs/mkdocs | chriskiehl/gooey | |
|---|---|---|---|
| Stars | 21,991 | 22,048 | 21,889 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 2/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires Python 3.14, uv, and at least one provider account or local model server configured via .env.
Free Claude Code is a proxy server that sits between Anthropic's Claude Code tool, the official command-line coding assistant, and a different model backend, so you can keep Claude Code's interface but have requests answered by an alternative provider, including free or local ones. The description sums it up as using Claude Code for free in the terminal, in a VS Code extension or in a Discord-style bot, with voice supported. The README says the proxy implements Claude Code's Anthropic Messages protocol and translates traffic to six provider backends: NVIDIA NIM, Kimi, OpenRouter, DeepSeek, and the local servers LM Studio, llama.cpp and Ollama. You set a MODEL value in a .env file using a "provider_id/model/name" format, and can route the Opus, Sonnet, Haiku and fallback tiers to different providers via MODEL_OPUS, MODEL_SONNET and MODEL_HAIKU. The proxy supports streaming, tool use, reasoning/thinking blocks, and a /v1/models endpoint so Claude Code's /model picker can list backend models when gateway model discovery is enabled. To run it you install uv and Python 3.14, clone the repo, edit .env, and launch the server with uvicorn on port 8082. Claude Code is then pointed at the proxy by setting ANTHROPIC_BASE_URL to the proxy and giving ANTHROPIC_AUTH_TOKEN any local value. Optional extras include a Discord or Telegram bot wrapper for remote coding sessions and voice-note transcription via local Whisper or NVIDIA NIM. You would use this if you want Claude Code's interface but cannot or do not want to pay Anthropic directly, or if you want to run a local model on your own machine for privacy or cost reasons.
A proxy server that lets you use Anthropic's Claude Code CLI with free or local AI model backends like OpenRouter, DeepSeek, and Ollama instead of paying Anthropic directly.
Mainly Python. The stack also includes Python, uvicorn, Whisper.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.