Claude Code Mux

OpenRouter met Claude Code Router. They had a baby.

Now your coding assistant can use GLM 4.6 for one task, Kimi K2 Thinking for another, and Minimax M2 for a third. All in the same session. When your primary provider goes down, it falls back to your backup automatically.

⚡️ Multi-model intelligence with provider resilience

A lightweight, Rust-powered proxy that provides intelligent model routing, provider failover, streaming support, and full Anthropic API compatibility for Claude Code.

Claude Code → Claude Code Mux → Multiple AI Providers
              (Anthropic API)    (OpenAI/Anthropic APIs + Streaming)

Table of Contents

• Key Features
• Installation
• Quick Start
• Screenshots
• Usage Guide
• Routing Logic
• Configuration Examples
• Supported Providers
• Advanced Features
• CLI Usage
• Troubleshooting
• FAQ
• Performance
• Why Choose Claude Code Mux?
• Documentation
• Changelog
• Contributing
• License

Key Features

🎯 Core Features

• ✨ Modern Admin UI - Beautiful web interface with auto-save and URL-based navigation
• 🔐 OAuth 2.0 Support - FREE access for Claude Pro/Max, ChatGPT Plus/Pro, and Google AI Pro/Ultra
• 🧠 Intelligent Routing - Auto-route by task type (websearch, reasoning, background, default)
• 🔄 Provider Failover - Automatic fallback to backup providers with priority-based routing
• 🌊 Streaming Support - Full Server-Sent Events (SSE) streaming for real-time responses
• 🌐 Multi-Provider Support - 18+ providers including OpenAI, Anthropic, Google Gemini/Vertex AI, Groq, ZenMux, etc.
• ⚡️ High Performance - ~5MB RAM, <1ms routing overhead (Rust powered)
• 🎯 Unified API - Full Anthropic Messages API compatibility

🚀 Advanced Features

• 🔀 Auto-mapping - Regex-based model name transformation before routing (e.g., transform all claude-* to default model)
• 🎯 Background Detection - Configurable regex patterns for background task detection
• 🤖 Multi-Agent Support - Dynamic model switching via CCM-SUBAGENT-MODEL tags
• 📊 Live Testing - Built-in test interface to verify routing and responses
• ⚙️ Centralized Settings - Dedicated Settings tab for regex pattern management

Screenshots

Overview Dashboard

Main dashboard with router configuration and provider management

Provider Management

Add and manage multiple AI providers with automatic format translation

Model Mappings with Fallback

Configure models with priority-based fallback routing

Router Configuration

Set up intelligent routing rules for different task types

Live Testing Interface

Test your configuration with live API requests and responses

Supported Providers

18+ AI providers with automatic format translation, streaming, and failover:

• Anthropic-compatible: Anthropic (API Key/OAuth), ZenMux, z.ai, Minimax, Kimi
• OpenAI-compatible: OpenAI, OpenRouter, Groq, Together, Fireworks, Deepinfra, Cerebras, Moonshot, Nebius, NovitaAI, Baseten
• Google AI: Gemini (OAuth/API Key), Vertex AI (GCP ADC)

Anthropic-Compatible (Native Format)

• Anthropic - Official Claude API provider (supports both API Key and OAuth)
• Anthropic (OAuth) - 🆓 FREE for Claude Pro/Max subscribers via OAuth 2.0
• ZenMux - Unified API gateway (Sunnyvale, CA)
• z.ai - China-based, GLM models
• Minimax - China-based, MiniMax-M2 model
• Kimi For Coding - Premium membership for Kimi

OpenAI-Compatible

• OpenAI - Official OpenAI API (supports both API Key and OAuth)
• OpenAI (OAuth) - 🆓 FREE for ChatGPT Plus/Pro subscribers via OAuth 2.0 (GPT-5.1, GPT-5.1 Codex)
• OpenRouter - Unified API gateway (500+ models)
• Groq - LPU inference (ultra-fast)
• Together AI - Open source model inference
• Fireworks AI - Fast inference platform
• Deepinfra - GPU inference
• Cerebras - Wafer-Scale Engine inference
• Moonshot AI - China-based, Kimi models (OpenAI-compatible)
• Nebius - AI inference platform
• NovitaAI - GPU cloud platform
• Baseten - ML deployment platform

Google AI

• Gemini - Google AI Studio/Code Assist API (supports both OAuth and API Key)
• Gemini (OAuth) - 🆓 FREE for Google AI Pro/Ultra subscribers via OAuth 2.0 (Code Assist API)
• Vertex AI - GCP platform with ADC authentication (supports Gemini, Claude, Llama via Model Garden)

Installation

Option 1: Download Pre-built Binaries (Recommended)

Download the latest release for your platform from GitHub Releases.

Linux (x86_64)

# Download and extract (glibc)
curl -L https://github.com/9j/claude-code-mux/releases/latest/download/ccm-linux-x86_64.tar.gz | tar xz

# Or download musl version (static linking, more portable)
curl -L https://github.com/9j/claude-code-mux/releases/latest/download/ccm-linux-x86_64-musl.tar.gz | tar xz

# Move to PATH
sudo mv ccm /usr/local/bin/

macOS (Intel)

# Download and extract
curl -L https://github.com/9j/claude-code-mux/releases/latest/download/ccm-macos-x86_64.tar.gz | tar xz

# Move to PATH
sudo mv ccm /usr/local/bin/

macOS (Apple Silicon)

# Download and extract
curl -L https://github.com/9j/claude-code-mux/releases/latest/download/ccm-macos-aarch64.tar.gz | tar xz

# Move to PATH
sudo mv ccm /usr/local/bin/

Windows

1. Download ccm-windows-x86_64.zip
2. Extract the ZIP file
3. Add the directory containing ccm.exe to your PATH

Verify Installation

ccm --version

Option 2: Install via Cargo

If you have Rust installed, you can install directly from crates.io:

cargo install claude-code-mux

This will download, compile, and install the ccm binary to your cargo bin directory (usually ~/.cargo/bin/).

Verify Installation

ccm --version

Option 3: Build from Source

Prerequisites

• Rust 1.70+ (install from rustup.rs)

Build Steps

# Clone the repository
git clone https://github.com/9j/claude-code-mux
cd claude-code-mux

# Build the release binary
cargo build --release

# The binary will be available at target/release/ccm

Install to PATH (Optional)

# Copy to /usr/local/bin for global access
sudo cp target/release/ccm /usr/local/bin/

# Or add to your shell profile (e.g., ~/.zshrc or ~/.bashrc)
export PATH="$PATH:/path/to/claude-code-mux/target/release"

Run Directly Without Installing (Optional)

# From the project directory
cargo run --release -- start

Quick Start

1. Start Claude Code Mux

ccm start

The server will start on http://127.0.0.1:13456 with a web-based admin UI.

💡 First-time users: A default configuration file will be automatically created at:

• Unix/Linux/macOS: ~/.claude-code-mux/config.toml
• Windows: %USERPROFILE%\.claude-code-mux\config.toml

2. Open Admin UI

Navigate to:

http://127.0.0.1:13456

You'll see a modern admin interface with these tabs:

• Overview - System status and configuration summary
• Providers - Manage API providers
• Models - Configure model mappings and fallbacks
• Router - Set up routing rules (auto-saves on change!)
• Test - Test your configuration with live requests

3. Configure Claude Code

Set Claude Code to use the proxy:

export ANTHROPIC_BASE_URL="http://127.0.0.1:13456"
export ANTHROPIC_API_KEY="any-string"
claude

That's it! Your setup is complete.

Usage Guide

Step 1: Add Providers

Navigate to Providers tab → Click "Add Provider"

Example: Add Anthropic with OAuth (🆓 FREE for Claude Pro/Max)

1. Select provider type: Anthropic
2. Enter provider name: claude-max
3. Select authentication: OAuth (Claude Pro/Max)
4. Click "🔐 Start OAuth Login"
5. Authorize in the popup window
6. Copy and paste the authorization code
7. Click "Complete Authentication"
8. Click "Add Provider"

💡 Pro Tip: Claude Pro/Max subscribers get unlimited API access for FREE via OAuth!

Example: Add ZenMux Provider

1. Select provider type: ZenMux
2. Enter provider name: zenmux
3. Select