Mcp Multi Model

What is Mcp Multi Model?

Mcp Multi Model is a single MCP server that gives Claude Code access to multiple LLM providers (OpenAI, Gemini, DeepSeek, Kimi, Grok, Perplexity, local models, and more) for image/video generation, web search, smart model routing, and side-by-side comparisons. It is designed for developers who want to use the best model for each task without switching tabs or managing separate integrations.

How to use Mcp Multi Model?

Install via npx mcp-multi-model or clone the repository and run npm install && npm run setup. Add the server to your Claude Code MCP config (e.g., ~/.mcp.json) with the required API keys in the env block. Optionally copy config.example.yaml to config.yaml to customize models, endpoints, and defaults.

Key features of Mcp Multi Model

• Smart routing – auto-picks the best model for the task
• Image generation via Gemini Imagen and OpenAI GPT Image
• Video generation via Gemini Veo 3.1
• Built-in web search (Google Search grounding, Kimi Chinese web)
• Compare models side by side with ask_all or ask_both
• One-line install with npx, no build step
• Supports local models (Ollama, LM Studio, llama.cpp, vLLM)
• Auto-retry, fallback, cost tracking, and response caching

Use cases of Mcp Multi Model

• Generate images and videos directly from the terminal without leaving Claude Code
• Research topics using web search grounded models (Gemini, Kimi)
• Compare answers from multiple models to choose the best solution
• Translate between Chinese and English using dedicated tools
• Write code cheaply by routing to DeepSeek or local models

FAQ from Mcp Multi Model

What models are supported?

12+ providers are preconfigured, including OpenAI, Gemini, DeepSeek, Kimi, Grok, Perplexity, Anthropic (via OpenRouter), Mistral, Groq, Qwen, GLM, Together, and local runners (Ollama, LM Studio, llama.cpp, vLLM). Models without an API key are skipped automatically.

How do I install Mcp Multi Model?

Add the server to your MCP config with "command": "npx", "args": ["-y", "mcp-multi-model"] and set your API keys in the env block. Alternatively, clone the repo and run npm install && npm run setup for an interactive wizard.

Do I need API keys?

Yes, for cloud models you need the corresponding API keys (e.g., DEEPSEEK_API_KEY, GEMINI_API_KEY). Local models (Ollama, LM Studio, etc.) require no API key and run entirely on your machine.

Is my data sent to the server author?

No. Mcp Multi Model is a local relay. No telemetry, analytics, or data is sent to the author. Prompts go directly from your machine to the LLM provider you configured.

How do I add a new model?

Add one block to your config.yaml file specifying the model name, adapter, endpoint, API key environment variable, and optional pricing/fallback settings. The server dynamically generates tools from your config.