Agentic Coding on Multi-SWE-Bench
44.3Pass@1Claude Sonnet-4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Sonnet-4.52026.03 | 44.3 | |
| Gemini 3-pro2026.03 | 42.7 | |
| Seed1.82026.03 | 42 | |
| GPT-5 High2026.03 | 41.7 | |
| Gemini 2.5-pro2026.03 | 20.7 |
| Method | Links | |
|---|---|---|
| Claude Sonnet-4.52026.03 | 44.3 | |
| Gemini 3-pro2026.03 | 42.7 | |
| Seed1.82026.03 | 42 | |
| GPT-5 High2026.03 | 41.7 | |
| Gemini 2.5-pro2026.03 | 20.7 |