Visual Grounding on HR-Bench 4K
96.2Accuracy (Single)LazyMCoT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LazyMCoTBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 96.2 | 59 | 77.6 | |
| HiDeBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 95.5 | 59.3 | 77.4 | |
| LazyMCoTBackbone=Qwen3-VL-8B-Instruct, Training Free=true2026.06 | 94.3 | 66.5 | 80.4 | |
| DeepEyesBackbone=Qwen2.5-VL-7B-Instruct, Training Free=false2026.06 | 91.3 | 59 | 75.1 | |
| Thyme-VLBackbone=Qwen2.5-VL-7B-Instruct, Training Free=false2026.06 | 91 | 63 | 77 | |
| Base (Qwen3-VL-8B-Instruct)Backbone=Qwen3-VL-8B-Instruct, Training Free=false2026.06 | 90.5 | 66.3 | 78.4 | |
| ViCropBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 90.5 | 57.5 | 74 | |
| DeepScanBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 90.1 | 59.7 | 75 | |
| LazyMCoTBackbone=InternVL3-8B-Instruct, Training Free=true2026.06 | 89.5 | 58.8 | 74.1 | |
| TreeVGRBackbone=Qwen2.5-VL-7B-Instruct, Training Free=false2026.06 | 89.5 | 61.5 | 72.7 | |
| DyfoBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 89.2 | 53.5 | 71.3 | |
| Base (Qwen2.5-VL-7B-Instruct)Backbone=Qwen2.5-VL-7B-Instruct, Training Free=false2026.06 | 88.8 | 54.8 | 71.8 | |
| HiDeBackbone=InternVL3-8B-Instruct, Training Free=true2026.06 | 88.5 | 57.5 | 73 | |
| ZoomRefineBackbone=Qwen2.5-VL-7B-Instruct, Training Free=true2026.06 | 88.5 | 55.3 | 71.5 | |
| ViCropBackbone=InternVL3-8B-Instruct, Training Free=true2026.06 | 88 | 57 | 72.5 | |
| Base (InternVL3-8B-Instruct)Backbone=InternVL3-8B-Instruct, Training Free=false2026.06 | 82.8 | 58.8 | 70.8 | |
| GPT-4oTraining Free=true2026.06 | 70 | 48 | 59 |