Ccrelay

CCRelay

Use third-party models in Claude Code, Claude Desktop, and the ChatGPT desktop app (ChatGPT Work and Codex).

CCRelay is a free, open-source desktop app for macOS and Windows. It lets Claude Code, Claude Desktop (third-party inference and Cowork), the ChatGPT desktop app (ChatGPT Work and Codex), and Codex CLI use third-party models such as GLM, Kimi, DeepSeek, Gemini, Qwen, MiniMax, and Xiaomi MiMo, or any OpenAI- or Anthropic-compatible API. Keep the client you already use and switch models from one place. A local proxy converts between Anthropic and OpenAI formats automatically. A VS Code extension is also available.

Download: Latest release — macOS .dmg (Apple Silicon, Intel) · Windows .exe (x64, arm64) · VS Code extension on Marketplace and Open VSX

Website: https://ccrelay.inflab.org · 中文文档 · 日本語 · 한국어

Table of Contents

- Supported clients
- Get started in 3 steps
- Core Features
- FAQ
- Privacy and local data
- Supported providers
- Installation
- Manual configuration
- Client Integrations
- Usage Guide
  - Multi-Instance Mode
  - Provider Modes
  - Model Mapping
  - Claude Desktop / Cowork Model ID Restrictions
  - OpenAI Format Conversion
  - Web UI Dashboard
- External web search
- Configuration
- API Endpoints
- Commands
- Development
- File Locations
- License

Supported clients

Details for each client: Client Integrations.

Get started in 3 steps

1. Install the desktop app. Download it from Releases and open it. The local proxy starts on http://127.0.0.1:7575.
2. Add a provider. Open Providers → Add provider, pick a preset (GLM, Xiaomi MiMo, DeepSeek, MiniMax, Gemini, Azure OpenAI, Meituan LongCat, Astraflow) or enter any base URL, paste your API key, run the built-in test, and create it.
3. Connect your client. Open Client configuration,…