Mathematical Problem Solving on AIME 2025 (Accuracy)
62.2AccuracyKACE
Evaluation results
| Method | Links | |
|---|---|---|
| KACESolver model=gpt-4.1-mini2026.05 | 62.2 | |
| Tiered + GEPASolver model=gpt-4.1-mini2026.05 | 56.6 | |
| Tiered + ACESolver model=gpt-4.1-mini2026.05 | 54.5 | |
| Tiered Best-of-NSolver model=gpt-4.1-mini2026.05 | 52.2 | |
| Best-of-5 self-consistencySolver model=gpt-4.1-mini2026.05 | 51.8 |