| Benchmark | Description | Gemini 3.5 Flash | Gemini 3 Flash | Gemini 3.1 Pro | Claude Sonnet 4.6 | Claude Opus 4.7 | GPT-5.5 | |
|---|---|---|---|---|---|---|---|---|
| Terminal-Bench 2.1 | Agentic terminal coding | Terminus-2 harness | 76.2% | 58.0% | 70.3% | — | 66.1% | 78.2% |
| SWE-Bench Pro (Public) | Diverse agentic coding tasks | Single attempt | 53.9% | 48.4% | 54.2% | 53.0% | 64.3% | 58.6% |
| MCP Atlas | Multi-step workflows using MCP | 83.6% | 62.0% | 78.2% | 69.5% | 79.1% | 75.3% | |
| Toolathlon | Real-world general tool use | 56.5% | 49.4% | — | — | — | 55.6% | |
| OSWorld-Verified | Agentic computer use | 78.4% | 65.1% | 76.2% | 72.5% | 78.0% | 78.7% | |
| Finance Agent v2 | Financial analysis and decision-making |
57.9% | 42.6% | 43.0% | 51.0% | 51.5% | 51.8% | |
| GDPval-AA | Economically valuable knowledge work |
Elo Rating | 1,656 | 1,204 | 1,314 | 1,674 | 1,753 | 1,773 |
| CharXiv Reasoning | Information synthesis from complex charts |
No tools | 84.2% | 80.3% | 83.3% | 70.5% | 82.1% | 84.1% |
| MMMU-Pro | Multimodal understanding and reasoning |
No tools | 83.6% | 81.2% | 80.5% | 74.5% | 75.2% | 81.2% |
| Blueprint-Bench 2 | Agentic spatial reasoning | Normalized score | 33.6% | 0.0% | 26.5% | 6.7% | 24.5% | 36.2% |
| MRCR v2 (8-needle) | Long context performance | 128k (average) 1M (pointwise) |
77.3% 26.6% |
67.2% 22.1% |
84.9% 26.3% |
84.9% not supported |
59.3% not supported |
94.8% not supported |
| Humanity’s Last Exam | Academic reasoning (full set, text + MM) |
40.2% | 33.7% | 44.4% | 33.2% | 46.9% | 41.4% | |
| ARC-AGI-2 | Abstract reasoning puzzles | 72.1% | 33.6% | 77.1% | 58.3% | 75.8% | 85.0% |