评测来源/OSWorld-Verified
观察中专业工作
OSWorld-Verified
XLANG Lab · OSWorld-Verified 的公开模型评测结果。
本来源的模型成绩
| Claude Fable 5 | 85 | 85.0 | 2026-06-09 |
| Claude Opus 4.8 | 83.4 | 83.4 | 2026-06-09 |
| Claude Sonnet 5 | 81.2 | 81.2 | 2026-06-30 |
| Muse Spark 1.1 | 80.8 | 80.8 | 2026-07-09 |
| GPT-5.5 | 78.7 | 78.7 | 2026-04-23 |
| Claude Sonnet 4.6 | 78.5 | 78.5 | 2026-06-30 |
| Claude Opus 4.7 | 78 | 78.0 | 2026-04-23 |
| Gemini 3.1 Pro Preview | 76.2 | 76.2 | 2026-05-19 |
| GPT-5.4 | 75 | 75.0 | 2026-04-23 |
| MiniMax-M3 | 70.06 | 70.1 | 2026-06-01 |