Dual Model MCP Server

A MCP (Model Context Protocol) server that queries Claude Sonnet 4.6 and OpenAI GPT-5.4 in parallel via OpenRouter and returns structured, multi-perspective responses.

The Problem

Sometimes a single AI model gets stuck in a particular perspective or reasoning pattern. You ask a question, get a good answer, but you know there's another angle, another approach that might be equally valuable (or better). Switching between different models, waiting for separate responses, losing context. It's tedious.

The Solution

Dual Model MCP Server sends your prompt to both Sonnet and GPT-5. simultaneously, giving you two independent, high-quality responses side-by-side. Compare, contrast, combine—all in one go. Perfect for:

• Decision-making: See technical/medical/business/research/legal questions from multiple angles
• Quality assurance: Spot blind spots in reasoning or missed edge cases
• Creative work: Get diverse perspectives on problems
• Validation: Cross-check facts and arguments between models

Features

• ⚡ Parallel queries – Both models respond simultaneously, not sequentially
• 📋 Structured responses – 6-8 concise paragraphs (analysis → context → evidence → arguments → alternatives → reflection → conclusion)
• 🔧 Easy integration – Works seamlessly with Cherry Studio, Claude Desktop, or any MCP client
• 🎯 Customizable system prompts – Use default structured prompt or define your own
• 🚀 Autostart support – Windows Task Scheduler integration for headless operation

Quick Start

Installation

git clone https://github.com/Firnschnee/dual-model-mcp.git
cd dual-model-mcp
npm install

Setup

1. Get an OpenRouter API key:

   • Go to openrouter.ai
   • Create an account / sign in
   • Copy your API key from settings

2. Create .env file:

   OPENROUTER_API_KEY=your_actual_api_key_here

3. Build & run:

   npm run build
   npm start

   You should see:

   ✅ Server läuft! Warte auf MCP-Anfragen via STDIO...

Usage

With Cherry Studio

1. Open Cherry Studio

2. Settings → MCP Servers → Add

3. Fill in:

   • Name: Dual Model MCP
   • Command: node
   • Arguments: C:\Users\[YourUsername]\dual-model-mcp\build\index.js
   • Working directory: C:\Users\[YourUsername]\dual-model-mcp

4. Save & restart Cherry Studio

5. Choose the MCP Server in the chat windows, ask a question & both models respond!

Autostart on Windows

Make the server start automatically on boot:

1. Open Task Scheduler (Win + R → taskschd.msc)

2. Create Basic Task

3. General:

   • Name: Dual Model MCP Server
   • ✅ Run with highest privileges

4. Trigger:

   • At startup

5. Action:

   • Program: C:\Program Files\nodejs\node.exe
   • Arguments: C:\Users\[YourUsername]\dual-model-mcp\build\index.js
   • Start in: C:\Users\[YourUsername]\dual-model-mcp

6. Finish

7. Test: Restart your PC, then check if server started

Stack & Dependencies

| Aspect | Technology | |--------|------------| | Language | TypeScript | | Protocol | Model Context Protocol (MCP) | | API | OpenRouter (supports 200+ models) | | Runtime | Node.js 18+ | | Build | tsc + npm |

Cost & Token Usage

Be aware, that this might cost a lot of tokens! max_tokens is currently set to 6000 to guarantee a deep dive analysis on almost any topic.

Customization

Use different models

Edit src/index.ts, line ~20:

const MODELS = {
  SONNET: "anthropic/claude-sonnet-4.6",
  GPT5: "openai/gpt-5.4",  // Change to any OpenRouter model
} as const;

Adjust response length

In src/index.ts, find queryModel():

const requestBody: OpenRouterRequest = {
  model,
  messages,
  temperature: 0.7,      // 0–1, lower = more consistent
  max_tokens: 6000,      // Increase for longer responses
};

Then rebuild:

npm run build

Contributing

Found a bug? Have an idea? Fork & submit a PR! 🚀

License

MIT License – See LICENSE file