| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Scibench fund | PoT + REFTOOL | Accuracy74.6 | 24 | 4mo ago | |
| UGPhysics | Accuracy67.6 | 12 | 1mo ago | ||
| PHYSICS (held-out) | DBS20.8 | 9 | 4mo ago | ||
| Interphyre Two Body Problem level 1.0 (test) | Qwen-2.5-7B HEXA | Accuracy34 | 4 | 25d ago | |
| Interphyre Down to Earth level 1.0 (test) | Qwen-2.5-7B HEXA | Accuracy72 | 4 | 25d ago | |
| Interphyre Catapult level 1.0 (test) | GPT-OSS-120B HEXA | Accuracy54 | 2 | 25d ago |
