SciAgentBench Gemini-2.5-Pro-Think (setting unspecified): benchmark results
Provider: Other.
Aggregate Rating 1667 ± 39, rank #512 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SciAgentBench (No Tools) | 28.9 | Success rate (%, 0-100) on all 259 tasks; without tools: cha | 87.5 |
| SciAgentBench - L1 (Up to 3 Steps) | 51 | Success rate (%, 0-100) on the tasks of difficulty L1 (3 ste | 71.9 |
| SciAgentBench - Life Sciences | 21.9 | Success rate (%, 0-100) on the 32 life sciences tasks; with | 62.5 |
| SciAgentBench - Physics | 33.3 | Success rate (%, 0-100) on the 109 physics tasks; with tools | 62.5 |
| SciAgentBench - L2 (4 to 7 Steps) | 28.8 | Success rate (%, 0-100) on the tasks of difficulty L2 (4 to | 56.2 |
| SciAgentBench | 28.8 | Success rate (%, 0-100) on all 259 tasks; with tools: a ReAc | 50 |
| SciAgentBench - Materials Science | 21.2 | Success rate (%, 0-100) on the 37 materials science tasks; w | 43.8 |
| SciAgentBench - L3 (8 or More Steps) | 14.5 | Success rate (%, 0-100) on the tasks of difficulty L3 (8 ste | 40.6 |
| SciAgentBench - Chemistry | 28.9 | Success rate (%, 0-100) on the 81 chemistry tasks; with tool | 31.2 |
Interactive version: theaggregate.ai/model?slug=sciagentbench-gemini-2-5-pro-think-setting-unspecified · How It Works · Data refreshed daily, snapshot 2026-11-08.