Coding Agent Capacity on SWE-Bench Multi
74ScoreClaude-Opus-4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude-Opus-4.52026.06 | 74 | |
| Gemini-3-pro High2026.06 | 72.7 | |
| Seed2.0 Pro2026.06 | 71.7 | |
| GPT-5.2 High2026.06 | 68.8 | |
| Claude-Sonnet-4.52026.06 | 64.1 | |
| Gemini-3-Flash High2026.06 | 59 | |
| Claude-Opus-4.52026.06 | 52.8 | |
| Gemini-3-pro High2026.06 | 50.2 | |
| GPT-5-mini High2026.06 | 49.3 | |
| GPT-5.2 High2026.06 | 47.7 | |
| Claude-Sonnet-4.52026.06 | 47.7 | |
| Seed2.0 Pro2026.06 | 45.2 | |
| Seed2.0 Lite2026.06 | 41.1 |