GUI Grounding on UI-I2E-Bench
87.3Click AccuracyUI-Ins-32B
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| UI-Ins-32BTraining Protocol=Fine-tuned2025.09 | 87.3 | — | — | — | — | — | — | |
| Mark-Grid ScaffoldTraining Protocol=Training-free2025.09 | 81.8 | — | — | — | — | — | — | |
| UI-Ins-7BTraining Protocol=Fine-tuned2025.09 | 81.1 | — | — | — | — | — | — | |
| UGround-V1-72BTraining Protocol=Fine-tuned2025.09 | 76.3 | — | — | — | — | — | — | |
| UI-TARS-72BTraining Protocol=Fine-tuned2025.09 | 73.7 | — | — | — | — | — | — | |
| UI-TARS-1.5-7BTraining Protocol=Fine-tuned2025.09 | 73.2 | — | — | — | — | — | — | |
| InfiGUI-R1-3BTraining Protocol=Fine-tuned2025.09 | 69.7 | — | — | — | — | — | — | |
| UI-R1-E-3BTraining Protocol=Fine-tuned2025.09 | 69.1 | — | — | — | — | — | — | |
| UI-TARS-7BTraining Protocol=Fine-tuned2025.09 | 61.4 | — | — | — | — | — | — | |
| OS-ATLAS-7BTraining Protocol=Fine-tuned2025.09 | 58.6 | — | — | — | — | — | — | |
| UGround-7BTraining Protocol=Fine-tuned2025.09 | 54.2 | — | — | — | — | — | — | |
| Aguvis-7BTraining Protocol=Fine-tuned2025.09 | 53.2 | — | — | — | — | — | — | |
| OS-ATLAS-4BTraining Protocol=Fine-tuned2025.09 | 44.3 | — | — | — | — | — | — | |
| ShowUI-2BTraining Protocol=Fine-tuned2025.09 | 41.5 | — | — | — | — | — | — | |
| SeeClick-7BTraining Protocol=Fine-tuned2025.09 | 26.4 | — | — | — | — | — | — | |
| Scaffold PromptingTraining Protocol=Training-free2025.09 | 25 | — | — | — | — | — | — | |
| Grid-Augmented VisionTraining Protocol=Training-free2025.09 | 9.5 | — | — | — | — | — | — | |
| AguvisSize=7B, Model Category=GUI-specific Models (SFT)2025.10 | — | 60.3 | 47.6 | 45.1 | 61.1 | 48.4 | 53.2 | |
| HyperClickSize=3B, Model Category=Ours2025.10 | — | 77.9 | 59 | 81 | 81.1 | 66.1 | 71.8 | |
| HyperClickSize=7B, Model Category=Ours2025.10 | — | 80.4 | 67.5 | 84.2 | 84.8 | 71.4 | 76.5 | |
| OmniParser (Wan et al., 2024)Model Category=GUI-specific Models (SFT)2025.10 | — | 67.6 | 45.5 | 30.8 | 54.3 | 52.4 | 53.1 | |
| OmniParser (Yu et al., 2025)Model Category=GUI-specific Models (SFT)2025.10 | — | 69.4 | 42.4 | 40.7 | 57 | 53.5 | 54.8 | |
| OS-AtlasSize=7B, Model Category=GUI-specific Models (SFT)2025.10 | — | 68.1 | 48.9 | 52.2 | 63.2 | 55.8 | 58.6 | |
| Qwen2.5-VLSize=7B, Model Category=General Models2025.10 | — | 61.7 | 41.6 | 56.9 | 58.4 | 51 | 53.8 | |
| SeeClickSize=9.6B, Model Category=GUI-specific Models (SFT)2025.10 | — | 37.2 | 15.8 | 18.2 | 37.1 | 19.9 | 26.4 | |
| ShowUISize=2B, Model Category=GUI-specific Models (SFT)2025.10 | — | 53.9 | 30.4 | 29.6 | 51.3 | 35.6 | 41.5 | |
| UGround-V1Size=7B, Model Category=GUI-specific Models (SFT)2025.10 | — | 73.5 | 65.7 | 70.8 | 81.3 | 63.6 | 70.3 | |
| UI-I2E-VLMSize=7B, Model Category=GUI-specific Models (SFT)2025.10 | — | 76.2 | 64 | 62.1 | 72 | 67.9 | 69.5 | |
| UI-R1Size=3B, Model Category=GUI-specific Models (RL)2025.10 | — | 67.8 | 46.2 | 58.1 | 67.9 | 52.8 | 58.5 | |
| UI-TARSSize=7B, Model Category=GUI-specific Models (SFT)2025.10 | — | 65.7 | 58 | 56.5 | 71.4 | 55.3 | 61.4 |