| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| red-wine benchmark | gpt-5.4 | Utility61.84 | 9 | 2mo ago | |
| AZR | PROPEL | Utility19.95 | 6 | 1mo ago | |
| Math | PROPEL | Utility Score41.93 | 4 | 1mo ago | |
| SWE OOD (held-out repositories) | PROPEL (OOD) | Utility19.59 | 2 | 1mo ago | |
| SWE RL-training repositories | PROPEL | Utility26.64 | 2 | 1mo ago |