List your company

Reasoning on GPQA (Error %, ErrorGap %, STP %)

10.82Error Rate (%)G-PAC
10.789210.997111.20511.4129Jan 30, 2026

Evaluation results

MethodLinks
G-PACScoring method=Verbalized score, Alpha=0.05, Epsilon=0.152026.0110.82019.6
PACScoring method=Verbalized score, Alpha=0.05, Epsilon=0.152026.0110.86017.07
PACScoring method=Logits-based score, Alpha=0.05, Epsilon=0.152026.0111.24308.73
G-PACScoring method=Router-based score, Alpha=0.05, Epsilon=0.152026.0111.34030.54
PACScoring method=Router-based score, Alpha=0.05, Epsilon=0.152026.0111.57034.21
G-PACScoring method=Logits-based score, Alpha=0.05, Epsilon=0.152026.0111.59013.78