| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| OSWorld | EvoCUA-32B | OS Success Rate75 | 45 | 1d ago | |
| OSWorld (Verified) | Score75 | 12 | 14d ago | ||
| AgentHijack (test) | AgentHijack-Agent | Clean Rate27.8 | 10 | 8d ago | |
| OpenComputer | Success Rate68.3 | 8 | 14d ago | ||
| OSWorld (test) | Success Rate42.5 | 6 | 3mo ago | ||
| WindowsAgentArena | ToolCUA-8B | Accuracy33.8 | 4 | 21d ago | |
| Windows Agent Arena (test) | Success Rate42.1 | 4 | 3mo ago | ||
| OSWorld | UI-TARS-2 | Headline SR47.5 | 3 | 8d ago | |
| OSWorld 50-step | Agent S2 | Headline SR34.5 | 2 | 8d ago | |
| VWA 910 | SGV on Gemini 2.5 | Headline SR54 | 1 | 8d ago | |
| ScreenSpot avg | Aguvis-72B | Headline Success Rate89.2 | 1 | 8d ago | |
| OSWorld 100-step | Agent S3 (bBoN) | Headline SR72.6 | 1 | 8d ago | |
| WAA | Agent S | Headline Success Rate18.2 | 1 | 8d ago | |
| AndroidW | OSCAR | Headline SR61.6 | 1 | 8d ago | |
| GAIA | OSCAR | Headline SR28.7 | 1 | 8d ago | |
| GAIA L1 | OS-Copilot/FRIDAY | Headline SR40.86 | 1 | 8d ago | |
| Mind2Web Live | Headline Success Rate51.1 | 1 | 8d ago | ||
| 643 live-web tasks | WebVoyager | Headline SR59.1 | 1 | 8d ago | |
| Mind2Web Online | - | - | 0 | 8d ago |