Mathematical Reasoning on Beyond-AIME VeRA-H VeRA-H Pro
58.34Avg@5 Accuracy (Seeds)Frontier Models (Mean)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Frontier Models (Mean)Model count=16, Seeds count=100, Verified variants count (H / Pro)=500 / 1002026.01 | 58.34 | 58.78 | 53.98 |