List your company

Mathematical Reasoning on AIME 24-26 (Pass@1 accuracy)

20Pass@1 AccuracyPUM
9.612.31517.7Jun 5, 2026

Evaluation results

MethodLinks
PUMSetting=Process-Only2026.0620
PUM+GRPOSetting=Outcome-Anchored2026.0617.78
Vanilla GRPOSetting=Outcome-Anchored2026.0616.67
PRM+GRPOSetting=Outcome-Anchored2026.0614.44
PQMSetting=Process-Only2026.0613.33
PQM+GRPOSetting=Outcome-Anchored2026.0610
PRMSetting=Process-Only2026.0610