Data

A fixed prompt matrix, re-run monthly against pinned model snapshots and published raw.

A fixed prompt matrix, re-run monthly against pinned model snapshots. Raw CSV and JSON, a written methodology, and a changelog. Prompts are never silently edited.

Drops

The first drop is in preparation. Each one ships the raw answers, the matrix version, the pinned snapshot identifiers, and the evaluation criteria used — so anyone can recompute the numbers rather than take them on trust.

No control group. Movement is not attributed to any single change — including ours.