| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| QWENCLAWBENCH | Overall Score59.5 | 18 | 1mo ago | ||
| SkillsBench 11 domains | Overall Score44.8 | 16 | 1mo ago | ||
| SkillsBench 1.0 (test) | EvoSkills (Claude Opus 4.6) | Pass Rate (With Skills)71.1 | 8 | 3mo ago | |
| AppWorld online mode | SkillComposer-4B | Accuracy57.7 | 3 | 1mo ago |