Short-term planning on GUI-Act-Web
94.54Type Success RateSSL-7B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SSL-7Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 94.54 | 89.25 | 87.23 | |
| RL-Continous-7Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 94.15 | 89.25 | 86.46 | |
| GUI-R1-7Bmode=Post-Training2026.01 | 90.85 | 88.06 | 80.31 | |
| RL-Binary-7Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 90.78 | 88.06 | 80.31 | |
| GUI-R1-3Bmode=Post-Training2026.01 | 89.86 | 87.42 | 76.31 | |
| QwenVL2.5-7B*mode=Post-Training, SFT=Mix-3K2026.01 | 87.66 | 84.77 | 79.89 | |
| Os-Atlas-7Bmode=Post-Training2026.01 | 86.95 | 75.61 | 57.02 | |
| QwenVL2.5-7Bmode=Zero Shot2026.01 | 86.59 | 84.39 | 78.63 | |
| RL-Continous-3Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 86.38 | 85.49 | 72.69 | |
| SSL-3Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 85.77 | 87.52 | 85.22 | |
| RL-Binary-3Bmode=Post-Training, training=standard RLVR on Mix-3K2026.01 | 82.84 | 83.19 | 75.15 | |
| Os-Atlas-4Bmode=Post-Training2026.01 | 79.22 | 58.57 | 42.62 | |
| GPT-4omode=Zero Shot2026.01 | 77.09 | 45.02 | 41.84 | |
| QwenVL2.5-3B*mode=Post-Training, SFT=Mix-3K2026.01 | 76.95 | 66.34 | 61.69 | |
| UI-R1-3Bmode=Post-Training2026.01 | 75.89 | 79.43 | 67.31 | |
| QwenVL2.5-3Bmode=Zero Shot2026.01 | 56.1 | 64.28 | 55.61 |