Proof Writing on IMO-GradingBench
4.33ScoreRQGM prover (specialist)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RQGM prover (specialist)Search Tokens=88.0M2026.06 | 4.33 | 61.7 | 48.3 | |
| Static IMO25 proverSearch Tokens=N/A2026.06 | 4.07 | 55 | 55 | |
| HGM-H proverSearch Tokens=21.5M2026.06 | 3.73 | 51.7 | 45 | |
| RQGM prover (generalist)Search Tokens=37.9M2026.06 | 3.73 | 51.7 | 45 |