Code Agent on SWE-Bench Verified
0.809ScoreClaude Opus 4.5
Evaluation Results
| Method | Links | |
|---|---|---|
| Claude Opus 4.5Framework=OpenHands2026.02 | 0.809 | |
| Claude-Opus-4.52026.06 | 0.809 | |
| GPT-5.2 (xhigh)Framework=OpenHands2026.02 | 0.8 | |
| GPT-5.2 High2026.06 | 0.8 | |
| Gemini-3-Flash High2026.06 | 0.78 | |
| GLM-5Framework=OpenHands2026.02 | 0.778 | |
| Claude Sonnet 4.5Variant=Sonnet 4.52026.01 | 0.772 | |
| Claude-Sonnet-4.52026.06 | 0.772 | |
| Kimi K2.5Framework=OpenHands2026.02 | 0.768 | |
| Seed2.0 Pro2026.06 | 0.765 | |
| Gemini-3.0 ProVariant=Pro2026.01 | 0.762 | |
| Gemini 3 ProFramework=OpenHands2026.02 | 0.762 | |
| Gemini-3-pro High2026.06 | 0.762 | |
| GPT-5 HighVariant=High2026.01 | 0.749 | |
| GLM-4.7Framework=OpenHands2026.02 | 0.738 | |
| Seed2.0 Lite2026.06 | 0.735 | |
| MiMo-V2-FlashVariant=Flash2026.01 | 0.734 | |
| DeepSeek-V3.2 ThinkingThinking Mode=true2026.01 | 0.731 | |
| DeepSeek-V3.2Framework=OpenHands2026.02 | 0.731 | |
| Kimi-K2 ThinkingThinking Mode=true2026.01 | 0.713 | |
| GPT-5-mini High2026.06 | 0.679 |