GUI Grounding on ScreenSpot (test)
90.6Element AccuracyTrifuse+GUI-AIMA
Evaluation Results
| Method | Links | |
|---|---|---|
| Trifuse+GUI-AIMAModel Size=7B, Training Protocol=Training-based (TB)2026.02 | 90.6 | |
| Trifuse+GUI-ActorModel Size=7B, Training Protocol=Training-based (TB)2026.02 | 90.5 | |
| MiMo-VL-Miloco-7B2025.12 | 89.8 | |
| Trifuse+GUI-ActorModel Size=3B, Training Protocol=Training-based (TB)2026.02 | 89.5 | |
| MiMo-VL-7B-SFT-25082025.12 | 89.5 | |
| GUI-ActorModel Size=7B, Training Protocol=Training-based (TB)2026.02 | 88.3 | |
| GUI-AIMAModel Size=3B, Training Protocol=Training-based (TB)2026.02 | 88.1 | |
| GUI-ActorModel Size=3B, Training Protocol=Training-based (TB)2026.02 | 86.5 | |
| TrifuseModel Size=7B, Training Protocol=Training-free (TF)2026.02 | 86.2 | |
| Qwen2.5-VL-7B2025.12 | 84.7 | |
| GRPO w. Two.Annotation Noise Level=0%2025.11 | 83.6 | |
| GRPO w. Max.Annotation Noise Level=0%2025.11 | 83 | |
| GRPO w. Max.Annotation Noise Level=20%2025.11 | 82.6 | |
| GRPO w. Two.Annotation Noise Level=20%2025.11 | 82.4 | |
| GRPOAnnotation Noise Level=0%2025.11 | 82.2 | |
| GRPOAnnotation Noise Level=20%2025.11 | 81.8 | |
| TrifuseModel Size=3B, Training Protocol=Training-free (TF)2026.02 | 81.1 | |
| GRPO w. Max.Annotation Noise Level=40%2025.11 | 81 | |
| GRPO w. Two.Annotation Noise Level=40%2025.11 | 80.6 | |
| GRPO w. Two.Annotation Noise Level=50%2025.11 | 80.2 | |
| GRPOAnnotation Noise Level=40%2025.11 | 79.8 | |
| InternVL3-8B2025.12 | 79.5 | |
| GRPO w. Two.Annotation Noise Level=60%2025.11 | 79.4 | |
| GRPO w. Min.Annotation Noise Level=20%2025.11 | 79 | |
| GRPO w. Min.Annotation Noise Level=0%2025.11 | 79 | |
| GRPO w. Max.Annotation Noise Level=50%2025.11 | 77.8 | |
| GRPO w. Min.Annotation Noise Level=40%2025.11 | 77.6 | |
| GRPO w. Min.Annotation Noise Level=60%2025.11 | 77.4 | |
| GRPO w. Min.Annotation Noise Level=50%2025.11 | 77.4 | |
| GRPO w. Two.Annotation Noise Level=80%2025.11 | 77 | |
| GRPO w. Max.Annotation Noise Level=60%2025.11 | 76.6 | |
| GRPOAnnotation Noise Level=50%2025.11 | 76.2 | |
| GRPOAnnotation Noise Level=60%2025.11 | 75.8 | |
| GRPO w. Two.Annotation Noise Level=100%2025.11 | 75.8 | |
| GRPO w. Min.Annotation Noise Level=80%2025.11 | 75.2 | |
| GRPO w. Max.Annotation Noise Level=80%2025.11 | 74.2 | |
| GRPO w. Max.Annotation Noise Level=100%2025.11 | 73.6 | |
| GRPO w. Min.Annotation Noise Level=100%2025.11 | 73.2 | |
| GRPOAnnotation Noise Level=80%2025.11 | 72 | |
| GRPOAnnotation Noise Level=100%2025.11 | 71 | |
| Base ModelAnnotation Noise Level=Base2025.11 | 70.6 | |
| TAGModel Size=8.5B, Training Protocol=Training-free (TF)2026.02 | 57.5 |