Computer UI Automation on CUA-World (test)
50.1Average ScoreGemini 3 Flash
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Gemini 3 FlashModel Category=Large Models2026.04 | 50.1 | 22.6 | |
| Kimi-K 2.5Model Category=Large Models2026.04 | 37.1 | 12.8 | |
| Ours (2B distilled)Model Category=Small Models, Model Size=2B, Distillation Status=Distilled2026.04 | 22.5 | 4.4 | |
| Qwen3-VL-4BModel Category=Small Models, Model Size=4B2026.04 | 19.3 | 3.9 | |
| Qwen3-VL-2BModel Category=Small Models, Model Size=2B2026.04 | 12.7 | 1.6 |