← Anthropic · Claude Individual model view Anthropic · Claude

Claude Sonnet 4.5

Appears in 6 benchmarksMean lean (confidence − pass rate): -0.023/6 benchmarks lean overconfident (prospective probe)

Positioning spread

0.000.250.500.751.00SQuAD (factual recall)MMLU-Pro (knowledge)LegalBench (legal reasoning)MathBench (competition math)OmniMath (advanced math)SciCode (scientific code)performanceconfidence (red gap = overconfident)
Performance vs. confidence for Claude Sonnet 4.5, per benchmark (prospective probe).
BenchmarkTask accConfidenceF₁Leans
SQuAD (factual recall)0.590.550.75-0.04 calibrated
MMLU-Pro (knowledge)0.700.790.82+0.09 overconfident
LegalBench (legal reasoning)0.860.730.79-0.13 cautious
MathBench (competition math)0.931.000.97+0.06 overconfident
OmniMath (advanced math)0.570.750.72+0.18 overconfident
SciCode (scientific code)0.570.290.43-0.29 cautious

In the full cloud

-3-3-2-2-1-100112233Performance z-score within benchmark/probe →Confidence z-score →
Claude Sonnet 4.5 conditions all other model/condition points equal relative confidence and pass rate

Pairwise footprint

Match accuracy controls for the performance base-rate gap
Claude Sonnet 4.5 pairs
18/ 171
Claude Sonnet 4.5 mean tau
+0.036
All-pairs mean
+0.037
Claude Sonnet 4.5 p<0.05
10(55.6%)
-1.0-0.50.00.51.0Pair signal: do confidence gaps rank performance gaps? (Kendall tau-b)
all model pairs (observed) base-rate-matched null calibration-preserving null Claude Sonnet 4.5 pair (filled = p<0.05) Claude Sonnet 4.5 mean all-pairs mean
vs Claude 3 Haiku →vs Claude 3.5 Sonnet →vs Claude Haiku 4.5 → Compare with anything →