UI Editing on UIEdit-Bench
4.94Edit CorrectnessUI2CodeN
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| UI2CodeNtraining_stage=RL2025.11 | 4.94 | 4.8 | 4.87 | |
| Claude-4-Sonnet2025.11 | 4.83 | 4.54 | 4.69 | |
| UI2CodeNtraining_stage=SFT2025.11 | 4.64 | 4.57 | 4.6 | |
| GPT-52025.11 | 4.63 | 4.46 | 4.54 | |
| Gemini-2.5-Pro2025.11 | 4.42 | 4.17 | 4.3 | |
| Qwen-2.5-VL-72B2025.11 | 3.53 | 3.3 | 3.41 |