⚡ Claude Code Gateway Switcher

A standalone, local API Gateway switcher and model browser built specifically for Claude Code. Easily switch between 12+ AI model providers (including NVIDIA NIM, OpenRouter, Agent Router, Aerolink, and Gemini) with a single click, directly writing configurations to your Claude settings file.

✨ Features

• One-Click Gateway Switcher: Automatically writes your selected provider config to ~/.claude/settings.json.
• Live Model Browser: Fetches live free (and paid) models dynamically from target endpoints, bypassing CORS using a local backend proxy.
• Connection Testing: Instantly test API key status and balance limits before applying configurations.
• Preconfigured Providers:
  • 🚀 Agent Router (Claude Opus 4.8, GPT-5, DeepSeek V3.1)
  • ✈️ Aerolink (Claude Opus 4.6, Claude Sonnet)
  • 🧠 FreeModel AI (Claude Opus, Claude Sonnet)
  • 🌐 OpenRouter (100+ models - DeepSeek, Llama, Qwen)
  • 🟢 NVIDIA NIM (Nemotron, DeepSeek, Llama, Qwen, Kimi, GLM)
  • ✦ Google Gemini (Gemini 2.0 Flash, 1.5 Pro)
  • ⚡ Groq (Ultra-fast Distill models)
  • 🧩 Cerebras (Wafer-scale speed Llama)
  • 🐙 GitHub Models (Free GPT-4o, Llama 405B, DeepSeek-R1)
  • 🌬️ Mistral (Mistral Large, Codestral)
  • 🤗 Hugging Face (Serverless router endpoints)
  • 🪂 Chutes AI (Decentralized inference)
  • ⚙️ Custom Gateway (Any Anthropic/OpenAI-compatible URL)
• Modern Swiss Typographic UI: A sleek, minimal design with gradients and JetBrains Mono fonts.

📦 Getting Started (Local Run)

1. Download & Install

Clone or download this repository to your computer, open a terminal in the folder, and install dependencies:

# Navigate to the folder
cd claude-gateway-switcher

# Install dependencies (only Express.js is required)
npm install

2. Start the Switcher Server

Run the local backend server:

npm run dev

You should see output similar to this:

  ⚡ Free Model Gateway Hub running at:
  → http://localhost:3847

  📁 Config path: /Users/username/.claude/settings.json

3. Switch Gateways

1. Open http://localhost:3847 in your browser.
2. Select your gateway (e.g., NVIDIA NIM, OpenRouter, Aerolink, etc.).
3. Enter your API Key (links to get free keys are provided directly on each card).
4. Hit "Test Connection" to verify if the key is valid and has balance.
5. Click "Apply to Claude Code" (only supported for Anthropic-compatible format gateways).
6. Restart Claude Code in your terminal:
   copy

   claude
   

7. That's it! Your Claude Code CLI is now using your selected provider. 🎉

🔧 Behind the Scenes

The local Express server hosts standard static files and handles backend operations:

• GET /api/config — Reads ~/.claude/settings.json.
• POST /api/config — Updates ~/.claude/settings.json with chosen API key, base URL, and model.
• GET /api/models — Queries model catalogs from provider endpoints.
• GET /api/test — Verifies connections to the selected provider.