Nerf Watch

Nerf Watch gives the same private, code-graded tests to nine your AI and GPT models every day: your AI models through your AI Code on a your AI Max plan, and GPT models through Codex on your AI Pro.

Each model is compared with its own recent baseline. This connector reads the published results through two read-only tools.

MCP connector

Server URL: https://tokenusage.dev/mcp/nerfwatch
Login: None
Protocol: MCP 2026-07-28
Instructions: Read-only Nerf Watch data from tokenusage.dev. returns the latest public model scores, previous runs, changes and baselines. returns one model's retained public score history, oldest first. Model ids are available from . Both tools share one 300-second read cache in each server instance. Missing days are missing, not zero. Scores and changes are rounded to two decimals. Comparisons are within one suite version. A lower score is a lower score on this suite, not proof that a provider changed a model. Runs are noisy, and a provider's defaults, the endpoint or the runner itself can change.

Tools

Reads data2

Skills

None in the store listings or on the server