registrycompare

Role then format against Ask when ambiguous

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

ask-when-ambiguous

v1 by @nils
90.3
84.6 to 94.0 · holdout 89.4
Intervals overlap. The 1.4 point gap is not significant at this sample size.
Metricrole-then-formatask-when-ambiguousBetter
Public score91.790.3left
Holdout score90.389.4left
Interval width8.59.4left
Cost per run$1.850$1.643right
Median latency2.4s1.5sright
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 ask-when-ambiguous \ --model claude-fable-5