← OpenAI · GPT Individual model view OpenAI · GPT

GPT-4o

Appears in 6 benchmarksMean lean (confidence − pass rate): +0.084/6 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 GPT-4o, per benchmark (prospective probe).
BenchmarkTask accConfidenceF₁Leans
SQuAD (factual recall)0.540.630.74+0.09 overconfident
MMLU-Pro (knowledge)0.470.870.66+0.40 overconfident
LegalBench (legal reasoning)0.830.450.58-0.38 cautious
MathBench (competition math)0.800.990.89+0.19 overconfident
OmniMath (advanced math)0.300.580.52+0.28 overconfident
SciCode (scientific code)0.510.390.56-0.12 cautious

In the full cloud

-3-3-2-2-1-100112233Performance z-score within benchmark/probe →Confidence z-score →
GPT-4o conditions all other model/condition points equal relative confidence and pass rate

Pairwise footprint

Match accuracy controls for the performance base-rate gap
GPT-4o pairs
18/ 171
GPT-4o mean tau
+0.041
All-pairs mean
+0.037
GPT-4o p<0.05
11(61.1%)
-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 GPT-4o pair (filled = p<0.05) GPT-4o mean all-pairs mean
vs GPT-4o mini →vs GPT-5.2 → Compare with anything →