Agent-to-User Interface Generation on A2UI-Bench
3.13L1 ScoreQwen3-30B-A3B-Instruct
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Qwen3-30B-A3B-InstructPrompt=w/ schema2026.05 | 3.13 | 2.13 | 2.09 | 2.92 | 2.64 | 2.51 | 2.57 | |
| GPT-4o-miniPrompt=w/ schema2026.05 | 3.45 | 2.27 | 2.15 | 2.78 | 2.28 | 2.06 | 2.5 | |
| Qwen3-235B-A22B-InstructPrompt=w/ schema2026.05 | 4 | 2.87 | 2.76 | 3.32 | 3.32 | 2.95 | 3.2 | |
| GPT-5.4Prompt=w/ schema2026.05 | 4.02 | 3.59 | 3.27 | 3.46 | 3.73 | 3.17 | 3.54 | |
| DeepSeek-V3.1Prompt=w/ schema2026.05 | 4.19 | 2.54 | 2.47 | 3.27 | 3.35 | 2.95 | 3.13 | |
| Gemini-3.1-ProPrompt=w/ schema2026.05 | 4.25 | 3.2 | 2.96 | 3.53 | 3.55 | 3.04 | 3.42 | |
| Macaron-A2UI-VentiPrompt=w/o schema2026.05 | 4.47 | 3.36 | 3.28 | 3.95 | 3.76 | 3.52 | 3.72 | |
| Macaron-A2UI-GrandePrompt=w/o schema2026.05 | 4.67 | 3.22 | 2.91 | 3.95 | 3.74 | 3.47 | 3.66 |