AndroidControl
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| GUI planning | AndroidControl Low | SR (%)95.05 | 40 | |
| trajectory-based action prediction | AndroidControl | Step Accuracy48.2 | 34 | |
| GUI planning | AndroidControl High | SR75.8 | 30 | |
| GUI Automation | AndroidControl High | Task Match (TM)83.7 | 27 | |
| GUI reasoning | AndroidControl Low | SR91.8 | 24 | |
| UI Navigation | AndroidControl (offline) | Step Success Rate79.1 | 23 | |
| Action Prediction | AndroidControl Low v2 | Pass@1 Step Accuracy88.9 | 22 | |
| Action Prediction | AndroidControl High v2 | Pass@1 Step Accuracy68.59 | 22 | |
| GUI Understanding | AndroidControl High | Task Match Rate (TM)83.7 | 22 | |
| Mobile Agent Evaluation | AndroidControl Low (test) | Task Success Rate93.7 | 22 | |
| Step Accuracy | AndroidControl High Level v2 | Pass@164.3 | 20 | |
| Step success rate | AndroidControl Task-Unseen | SSR62.3 | 20 | |
| GUI Interaction Control | AndroidControl (High) | SR79.37 | 20 | |
| GUI Navigation | AndroidControl High | SR (Success Rate)76.3 | 17 | |
| General Agent Capability | AndroidControl High | Type Rate84.7 | 17 | |
| High-level instruction following | AndroidControl | Step Accuracy72.7 | 16 | |
| Short-term planning | AndroidControl Low | Type85.17 | 16 | |
| General Agent Capability | AndroidControl Low | Type Score97.2 | 15 | |
| Step Execution | AndroidControl High | Step Success Rate75.3 | 15 | |
| Mobile Agent Navigation | AndroidControl 1.0 (test) | Step Success Rate (SR)72.9 | 15 | |
| Mobile GUI Agent Execution | AndroidControl Curated Easy | Type Success Rate88.6 | 15 | |
| Mobile GUI Agent Execution | AndroidControl Curated-Hard | Type Rate80.8 | 15 | |
| Mobile UI Automation | AndroidControl High (test) | Success Rate (SR)80.3 | 14 | |
| Mobile navigation | AndroidControl | Step Accuracy76.1 | 14 | |
| Long-term Planning | AndroidControl High | Type Rate71.79 | 14 |
Showing 25 of 51 rows