/mcp

ModelPicker, inside your IDE.

Install the MCP server and get benchmark-backed model recommendations from inside Claude Code, Cursor, and any MCP-compatible client. No API key required.

Claude Code~/.claude/settings.json

{
  "mcpServers": {
    "modelpicker": {
      "type": "sse",
      "url": "https://modelpicker-mcp.onrender.com/sse"
    }
  }
}

CursorSettings → MCP → Add Server → SSE

URL: https://modelpicker-mcp.onrender.com/sse

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "modelpicker": {
      "type": "sse",
      "url": "https://modelpicker-mcp.onrender.com/sse"
    }
  }
}

Ask it things like…

RecommendI need a model for a customer support chatbot. Budget is $5/MTok max. What do you recommend?

CompareCompare Claude Sonnet 4.6 vs GPT-4o for structured JSON output — scores, pricing, tradeoffs.

FilterShow me all models under $1/MTok with at least 200K context window, sorted by coding score.

AlternativesI'm using GPT-4o but it's too expensive. Find me cheaper alternatives that score within 10% on tool calling.

Best for taskWhat's the best model for math and reasoning? I don't care about cost, just quality.

Benchmark dataList all benchmarks you test. What does the 'faithfulness' benchmark actually measure?

Value pickWhat's the best bang-for-buck model right now? Sort by quality per dollar.

Multi-criteriaI need a model that's good at coding AND structured output AND costs under $3/MTok. What are my options?