| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Scientific Agent Task Completion | ScienceAgentBench | Success Rate (SR)43.1 | 40 | |
| Scientific Code Generation | ScienceAgentBench (test) | SR32.8 | 32 | |
| Scientific Discovery | ScienceAgentBench | Score0.225 | 24 | |
| Defect Detection | ScienceAgentBench 12 confirmed defects, 102 tasks | Recall Average (RecA)100 | 12 | |
| Scientific Code Generation | ScienceAgentBench | SR25.5 | 10 | |
| Scientific Agent Task | ScienceAgentBench (test) | Success Rate (SR)18.6 | 6 | |
| Scientific-coding | ScienceAgentBench deterministic-eval 2024 | Pass Rate47.4 | 5 |