| Benchmark | Model | Rank | Results |
|---|---|---|---|
| code-generation-on-swe-bench-pro | Gemini 3.1 Pro | – | Resolved: 54.2 |
| code-generation-on-swe-bench-pro | gemini-3.1-pro (thinking) | – | Resolved: 46.10 |
| code-generation-on-terminal-bench-2-1 | Terminus-2 + Gemini 3.1 Pro | – | Accuracy: 74.0 |