Web Task Execution on WebVoyager and WebShop representative subset of 300+ tasks
52AccuracyAccio
Evaluation results
| Method | Links | |
|---|---|---|
| AccioAgent Backend=AgentOccam2026.05 | 52 | |
| AgentOccamMode=Default2026.05 | 49.6 | |
| AccioAgent Backend=BrowserUse2026.05 | 45.6 | |
| BrowserUseMode=Default2026.05 | 45 | |
| AccioAgent Backend=WebVoyager2026.05 | 40.6 | |
| WebVoyagerMode=Default2026.05 | 37.6 |