Diff cards · the flagship artifact

When a build changes, the record shows exactly what moved.

When a pinned build is succeeded, the record measures both sides of the succession on the scheduled battery, and every reading is dated and chain-verified. Builds move most in the days right after a launch, which is why an accelerated cadence is planned: a battery within 24 hours of a new build, and a daily rerun through the first week. Neither is built, so the succession is measured at the weekly cadence like everything else, and the record does not yet resolve that first week at daily resolution.

Published diffs

None yet, the first diff card publishes on the first roster succession after the public series begins. Cross-build comparisons are diffs, never drift: a successor is a different artifact and is labeled accordingly.

What a diff card looks like

frozen format, numbers and deltas only

IllustrativeDIFF #041 · [seat-id] · build 2026-XX-XX → build 2026-YY-YY Battery v1.0 · n=10 trials × 2 paraphrases · same protocol both sides CHANGED (passed effect-size floor; confirming re-measurement not yet built): B2 open-weights release 3.1 → 4.2 (Δ +1.1, q=0.008) A2 police facial recognition refusals 0/20 → 6/20 UNCHANGED: 21 of 25 questions within noise. Evidence: raw transcripts · seeds · replay command · chain hash a3f9…

No adjectives, no speculation. A clearly separated, human-gated “Notes” block may accompany a card; the card itself is generated by the pipeline and gated by the banned-adjective lint.

Depend on one of these models? Get the diff in your inbox the day it publishes, with the blast radius scoped to your builds.

Watch your builds →