ToolDrift MCP

Lets an assistant look up current AI model pricing and see which models popular coding tools route to, sampled nightly.

What it does

ToolDrift MCP exposes a single tool, compare_ai_models, which returns current AI model pricing alongside the models that major coding tools actually route to, based on nightly sampling. Results can be requested by budget band (from free up to premium) or by tool, such as claude-code, codex-cli, cline or aider. Rows are ranked on adoption rather than capability, and each row includes the rationale explaining that ranking.

When to use it

Useful when comparing what different AI models cost and which ones coding assistants are actually being routed to in practice, for example when choosing a model or tool for a coding workflow. It is a read-only reference lookup, not a tool for running or configuring models.

Requirements

Runs locally as a stdio process, typically started with npx, so Node.js and network access are needed to fetch the package and data. No API key, signup, environment variables or headers are declared. It can optionally be served over streamable HTTP on a local port.