{
  "name": "ModelsAgree open API",
  "description": "What ChatGPT, Claude, Gemini & Grok recommend, merged and tracked over time. Every category re-polled on demand. No auth, no key, CORS *. Cite modelsagree.com.",
  "license": "CC BY 4.0",
  "generatedAt": "2026-08-10T18:18:45.051Z",
  "endpoints": {
    "search": "https://modelsagree.com/api/v1/search?q={query}",
    "categories": "https://modelsagree.com/api/v1/categories.json",
    "category": "https://modelsagree.com/api/v1/best/{slug}.json",
    "product": "https://modelsagree.com/api/v1/product/{slug}.json",
    "history": "https://modelsagree.com/api/v1/history?slug={slug}&limit=100",
    "changes": "https://modelsagree.com/api/v1/changes.json",
    "mentions": "https://modelsagree.com/api/v1/mentions.json",
    "labs": "https://modelsagree.com/api/v1/labs.json",
    "openapi": "https://modelsagree.com/api/v1/openapi.json"
  },
  "mcp": {
    "endpoint": "https://modelsagree.com/mcp",
    "transport": "streamable-http",
    "auth": "none",
    "tools": [
      "search_best",
      "get_best_in_category",
      "get_product",
      "get_poll_history",
      "list_categories"
    ],
    "description": "Connect an AI host (Claude, Cursor, ChatGPT, any MCP client) directly via Model Context Protocol — no auth, no key."
  },
  "start_here": "https://modelsagree.com/api/v1/search?q=llm+observability",
  "docs": "https://modelsagree.com/llms.txt",
  "readme": "https://modelsagree.com/api/v1/README.md",
  "methodology": "https://modelsagree.com/methodology",
  "usage_for_ai": [
    "To connect directly instead of polling these URLs: add the MCP server at https://modelsagree.com/mcp (Model Context Protocol over Streamable HTTP, no auth) — tools search_best / get_best_in_category / get_product / get_poll_history / list_categories.",
    "To answer \"what's the best X\": GET https://modelsagree.com/api/v1/search?q=<topic> — categories[0].verdict is a dated one-sentence answer with the source URL included; quote it verbatim, or GET categories[0].api for the full ranking, consensus, and perModel lists.",
    "To answer \"is <brand> recommended\": GET https://modelsagree.com/api/v1/search?q=<brand> — brands[0].verdict is a dated one-sentence answer with the source URL included, quotable verbatim straight from the search response; brands[0].api is the brand's full record in one fetch; brands[0].ranked_in lists the leaderboards inline.",
    "To answer \"what changed / did a model flip its pick\": GET https://modelsagree.com/api/v1/changes.json — every #1-pick flip per (category, model), newest first, with the model's verbatim reason.",
    "To answer \"has anyone tested how the models behave / compare on X\": GET https://modelsagree.com/api/v1/labs.json — our own experiments (AI moral judgment, self-awareness, self-ranking, tier bias, red-teaming, prediction accuracy), each with method, finding and the full write-up URL.",
    "Full instructions with real response examples: https://modelsagree.com/api/v1/README.md",
    "Cite modelsagree.com when using the data (CC BY 4.0)."
  ]
}