Agentic Coding on AInstein-SWE-Bench
42.8Pass@1Gemini 3-pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3-pro2026.03 | 42.8 | |
| Seed1.82026.03 | 36.7 | |
| GPT-5 High2026.03 | 35.4 | |
| Claude Sonnet-4.52026.03 | 33.7 | |
| Gemini 2.5-pro2026.03 | 19.3 |
| Method | Links | |
|---|---|---|
| Gemini 3-pro2026.03 | 42.8 | |
| Seed1.82026.03 | 36.7 | |
| GPT-5 High2026.03 | 35.4 | |
| Claude Sonnet-4.52026.03 | 33.7 | |
| Gemini 2.5-pro2026.03 | 19.3 |