Mobile application exploration on ScaleWoB 10 apps (test)
20.9Average RewardReAct-vision
Evaluation Results
| Method | Links | |
|---|---|---|
| ReAct-visionModel=Gemini 3.1 Flash-Lite, Memory=Full history2026.06 | 20.9 | |
| JAMELModel=JAMEL-9B, Memory=Latent memory2026.06 | 20.7 | |
| ReAct-textModel=Gemini 3.1 Flash-Lite, Memory=Full history2026.06 | 19.9 | |
| MAI-UIModel=MAI-UI-8B, Memory=Device/cloud routing2026.06 | 8.4 | |
| Mobile-Agent-v3.5Model=GUI-Owl-1.5-8B, Memory=Sliding window + Notetaker2026.06 | 5.9 |