Autonomous Scientific Research on ARC-Bench experiment-stage CD:CE:RA = 25:25:50 (25 topics)
96.8Code Development ScoreAutoResearchClaw
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| AutoResearchClawMode=CoPilot2026.05 | 96.8 | 57.8 | 52.3 | 64.8 | |
| AIDE-ML2026.05 | 95.8 | 41.5 | 33.6 | 51.1 | |
| AutoResearchClawMode=Full-Auto2026.05 | 93.8 | 56.2 | 44.2 | 59.6 | |
| AI Scientist v22026.05 | 71.2 | 44.2 | 26.1 | 41.9 |