Compare models from your agent

Connect ModelCompare to Claude Code, Cursor, Codex, ChatGPT or any app that supports MCP. Your agent can look up current models and prices, run the same prompt on several models at once, and hand you a link to the side-by-side view. Runs use your ModelCompare credits at the same prices as the site.

Server URL

https://www.modelcompare.ai/mcp

Your app opens a ModelCompare page to sign in with Google the first time. Nothing to copy besides the URL.

Set up your app

Claude Code

Add the server, then run /mcp in Claude Code and pick modelcompare to sign in.

claude mcp add --transport http modelcompare https://www.modelcompare.ai/mcp

Cursor

Add this to ~/.cursor/mcp.json (or Settings → MCP → Add server), then click Connect.

{
  "mcpServers": {
    "modelcompare": {
      "url": "https://www.modelcompare.ai/mcp"
    }
  }
}

Codex CLI

Add the server, then sign in.

codex mcp add modelcompare --url https://www.modelcompare.ai/mcp
codex mcp login modelcompare

VS Code (Copilot)

Add the server, then start it from the MCP servers list and sign in.

code --add-mcp '{"name":"modelcompare","type":"http","url":"https://www.modelcompare.ai/mcp"}'

Gemini CLI

Add the server, then run /mcp auth modelcompare.

gemini mcp add --transport http modelcompare https://www.modelcompare.ai/mcp

Claude.ai and Claude Desktop

Settings → Connectors → Add custom connector, paste the URL, then Connect.

https://www.modelcompare.ai/mcp

ChatGPT

Turn on developer mode (Settings → Apps → Advanced), create an app with this URL and OAuth sign-in.

https://www.modelcompare.ai/mcp

Any other MCP client works with the URL above. Clients that can't sign in through the browser can use an API key instead.

Things to ask

Which model should I use for the summarizer in src/summarize.ts? Test it on 3 real inputs and pick the cheapest one that's good enough.
Ask GPT, Claude and Gemini this question and show me where they disagree.
What's the cheapest model with an intelligence score above 50?
How much would this prompt cost per 1,000 calls on each model?

Tools

list_models: Current models with price, intelligence score and measured speed.
estimate_cost: What a prompt would cost on each model, without running it.
compare: Runs a prompt (or up to 10 test inputs) on up to 6 models and returns every answer with cost, time and a link.
get_comparison: Reads back an earlier comparison with the full answers.

Comparisons run from an agent show up in your history on modelcompare.ai. A call is refused when its worst-case cost is over its max_cost_usd ($0.25 unless the agent sets it).