Cc Open Models Routing

cc-open-models-routing

Local HTTP proxy + CLI tool for routing Claude Code between **Z.AI (GLM-5.1)** and **MiniMax (M2.7)** based on model tier — giving you GLM quality for orchestration and MiniMax speed/cost for subagents.

Profiles

Profile	Provider	Main Model	Notes
glm-5.1	Z.AI (direct)	GLM-5.1	All tasks → Z.AI
minimax-m2.7	MiniMax (direct)	MiniMax-M2.7	All tasks → MiniMax
efficiency	Local proxy	GLM-5.1	Opus tier → Z.AI · Sonnet/Haiku tier → MiniMax

How Efficiency Mode Works

Claude Code
    │
    ▼ (ANTHROPIC_BASE_URL = http://localhost:3472)
┌─────────────────────────────────────┐
│         claude-proxy (port 3472)     │
│                                      │
│  POST /v1/messages                  │
│  { "model": "glm-5.1", ... }      │
│            ↓ model.startsWith("MiniMax")?
│       ┌────┴────┐                   │
│       no       yes                  │
│       ▼         ▼                   │
│   Z.AI     MiniMax                  │
│  api.z.ai  api.minimax.io          │
└─────────────────────────────────────┘

Prerequisites

• Bun runtime
• Windows (PowerShell-based env var management)
• Z.AI token plan subscription
• MiniMax token plan subscription

Setup

1. Set API Tokens

# Z.AI token (from z.ai dashboard)
[System.Environment]::SetEnvironmentVariable('CLAUDE_PROFILE_ZAI_TOKEN', 'your-zai-token', 'User')

# MiniMax token (from platform.minimax.io)
[System.Environment]::SetEnvironmentVariable('CLAUDE_PROFILE_MINIMAX_TOKEN', 'your-minimax-token', 'User')

Restart your terminal after setting.

2. Install

# Clone the repo
git clone https://github.com/iskisraell/cc-open-models-routing.git
cd cc-open-models-routing

# Install dependencies
bun install

3. One-Time Profile Setup

Choose the profile you want. For mixed routing (recommended):

# Apply efficiency mode — does NOT start Claude, just configure