Mobile UI Control on Robustness Benchmark
43.3LROS-Atlas-7B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OS-Atlas-7BCategory=Open-source 7B/8B2026.04 | 43.3 | 10.7 | |
| Qwen2.5-VL-3BCategory=Open-source 3B2026.04 | 30 | 35 | |
| UI-R1-3BCategory=Open-source 3B2026.04 | 29.5 | 29 | |
| VeriGUI-3BCategory=Ours2026.04 | 24.3 | 51.1 | |
| Qwen2.5-VL-7BCategory=Open-source 7B/8B2026.04 | 23.5 | 34.1 | |
| UI-S1-7BCategory=Open-source 7B/8B2026.04 | 20.5 | 43.6 | |
| VeriGUI-7BCategory=Ours2026.04 | 15.6 | 52.5 | |
| Gemini-3-flashCategory=Closed-source2026.04 | 15 | 37 | |
| UI-TARS-7BCategory=Open-source 7B/8B2026.04 | 13.4 | 45.5 | |
| GPT-5.1Category=Closed-source2026.04 | 11 | 21.9 |