评测来源/SWE-Atlas Codebase QnA
观察中编程
SWE-Atlas Codebase QnA
Scale AI · SWE-Atlas Codebase QnA 的公开模型评测结果。
本来源的模型成绩
| Claude Opus 4.7 | 81 | 81.0 | 2026-08-05 |
| GPT-5.5 | 80.8 | 80.8 | 2026-08-05 |
| GPT-5.5 | 79.1 | 79.1 | 2026-08-05 |
| Claude Opus 4.7 | 78.4 | 78.4 | 2026-08-05 |
| GPT-5.5 | 75 | 75.0 | 2026-08-05 |
| GLM-5.1 | 73.2 | 73.2 | 2026-08-05 |
| GPT-5.4 | 72.9 | 72.9 | 2026-08-05 |
| GPT-5.4 | 72.4 | 72.4 | 2026-08-05 |
| Claude Opus 4.6 | 71.9 | 71.9 | 2026-08-05 |
| Claude Opus 4.7 | 71.7 | 71.7 | 2026-08-05 |
| Claude Sonnet 4.6 | 70.3 | 70.3 | 2026-08-05 |
| DeepSeek V4 Pro | 67.8 | 67.8 | 2026-08-05 |
| Kimi K2.6 | 59.8 | 59.8 | 2026-08-05 |
| Gemini 3.1 Pro Preview | 45.6 | 45.6 | 2026-08-05 |
| GPT-5.5 | 45.43 | 45.4 | 2026-08-05 |
| GPT-5.4 | 40.8 | 40.8 | 2026-08-05 |
| GPT-5.4 | 36.3 | 36.3 | 2026-08-05 |
| Claude Opus 4.6 | 33.3 | 33.3 | 2026-08-05 |
| Claude Sonnet 4.6 | 31.2 | 31.2 | 2026-08-05 |
| Claude Opus 4.6 | 30 | 30.0 | 2026-08-05 |
| GLM-5 | 20.5 | 20.5 | 2026-08-05 |
| Gemini 3.1 Pro Preview | 13.5 | 13.5 | 2026-08-05 |
| Kimi K2.5 | 13.1 | 13.1 | 2026-08-05 |
| MiniMax-M2.5 | 10.3 | 10.3 | 2026-08-05 |
| Gemini 3 Flash Preview | 8.2 | 8.2 | 2026-08-05 |