GUI Task Execution on SPA-Bench single-app English Level 3
53.7Task Success RateGUI-explorer
Evaluation results
| Method | Links | |
|---|---|---|
| GUI-explorerInput=SOM, Base Model=GPT-4o2025.05 | 53.7 | |
| M3AInput=SOM, Base Model=GPT-4o2025.05 | 42 | |
| MobileAgentV2Input=SoM, Base Model=GPT-4o2025.05 | 20 | |
| AppAgentInput=SOM, Base Model=GPT-4o2025.05 | 14 | |
| AutoDroidInput=a11y tree, Base Model=GPT-4o2025.05 | 12 | |
| SeeActInput=SOM, Base Model=GPT-4o2025.05 | 12 | |
| CogAgentInput=screen, Base Model=CogAgent2025.05 | 0 | |
| DigiRLInput=screen, Base Model=DigiRL2025.05 | 0 |