Free Claude Code

Free Claude Code lets developers use coding agents with local models or a choice of cloud providers. It connects existing clients to a shared model catalog rather than requiring a separate coding workflow.

• Local model backends: Use Ollama, LM Studio or llama.cpp for inference on your own hardware, with models that support tools and enough context for agent requests.
• Coding agent connections: Connect Claude Code, Codex, OpenCode, Cline or Aider, with access through VS Code, JetBrains and desktop clients.
• Browser and phone access: Run Codex sessions in the browser, including background work and model changes within a session, or interact with agents through Discord and Telegram.
• Automatic fallback: After retries fail, the proxy tries another configured model across connected clients; a request may consume usage from multiple providers.
• Compatible agent features: Supported routes preserve streaming, tool calls, images and interleaved thinking, with separate model choices for Claude Code tiers and reasoning controls where providers support them.

The proxy runs on macOS, Windows and Linux. Local backends keep model inference on your machine; cloud routes send requests to the selected provider and need internet access. Voice input uses local Whisper or cloud transcription through NVIDIA NIM. Optional RTK filtering reduces terminal output sent to models. The project is independent of Anthropic.