List your company

Mathematical Reasoning on OlympiadBench (Acc, PCE, ECE)

57AccuracyC^2GSPG
48.6850.845355.16Apr 25, 2026

Evaluation results

MethodLinks
C^2GSPGConf.=logits2026.04570.2330.235
RLCRConf.=verbal2026.0456.80.2050.205
GRPOConf.=probe2026.0456.60.0140.096
RLCMConf.=probe2026.04560.0020.068
BaseConf.=probe2026.04490.0310.059