For agents

MCP server.

Ask your agent where a model is sold for less and which venue answered first. Two read-only tools over the list prices and probe timings this website publishes. No account and no key.

Connect

The server is at https://data.kottos.ai/mcp, over Streamable HTTP. Add it once:

# Claude Code; add --scope user to use it in every project
claude mcp add --transport http kottos https://data.kottos.ai/mcp

# Codex
codex mcp add kottos --url https://data.kottos.ai/mcp

In Cursor, add it to ~/.cursor/mcp.json:

{ "mcpServers": { "kottos": { "url": "https://data.kottos.ai/mcp" } } }

In Claude.ai and Claude Desktop, open Customize, then Connectors, and add a custom connector with the same URL.

Tools

  • list_models takes no input. It returns every model with a list price, how many venue and service tier offers sell it, and how many of those offers the probe measures.
  • compare_venues takes one model, by its name in list_models or by an API id such as claude-sonnet-5-5. For each venue and tier that sells it, it returns list prices in US dollars per million tokens (input, output, cache read, cache write for five minutes and for an hour, and long-context prices where they differ) and the probe's time to first token over the last hour: median, 90th percentile, samples, failures and the day's typical median. Measured offers come first, fastest first, and the rest follow, cheapest first.

Ask it

  • Which venue serves Claude Sonnet 5.5 fastest right now, and what does it cost there?
  • Compare GPT-6.1 Sol's standard, flex and ultrafast tiers on price and time to first token.
  • Where is DeepSeek V4 Pro cheapest, and is the cheapest venue slower?
  • Which models can I buy from more than one venue?

What the numbers are

  • Prices are list prices collected from providers' public pages. They can be out of date or wrong, and your contract can differ, so check the provider's page before you buy.
  • The probe sends one short, fixed prompt to each measured offer once a minute from Ashburn, Virginia, through the Kottos AI gateway. Its time to first token is mostly the network trip and the provider's queue, which is less than what a long prompt waits. Not every offer is probed, and the tool says which are.

Limits

  • 120 requests a minute from one address.
  • Agents and command-line clients connect directly. A web page on another site cannot call the server from a browser.
  • Protocol versions 2026-07-28, 2025-11-25, 2025-06-18 and 2025-03-26. Each request gets one JSON reply, and there is no session.

Privacy

We count requests per day by client name, method, tool and the model matched. We store no IP address and nothing else a request sends. The privacy policy has the details.

Questions

Kottos AI, Inc. · [email protected]