registrycompare

Role then format against First tool guess

Both scored on the same cases against claude-fable-5, same judge. The intervals tell you whether the gap is real or noise.

suite reasoning-chainmodel claude-fable-5n=168 public cases each

role-then-format

v2 by @ilya
91.7
86.5 to 95.0 · holdout 90.3

first-tool-guess

v1 by @system
75.3
68.0 to 81.5 · holdout 74.2
Intervals do not overlap. The 16.4 point gap is significant at this sample size.role-then-format wins
Metricrole-then-formatfirst-tool-guessBetter
Public score91.775.3left
Holdout score90.374.2left
Interval width8.513.5left
Cost per run$1.850$1.814right
Median latency2.4s1.4sright
Sample size168154left

Reading this verdict

Overlap
Two entries are only meaningfully apart when their confidence intervals do not overlap. A higher number with an overlapping interval is a coin flip, not a win.
Holdout
The public score can be gamed by tuning to visible cases. A gap between public and holdout is the tell.
Reproduce
evalness compare \ role-then-format first-tool-guess \ --model claude-fable-5