Fine-grained Visual Perception on V* Bench
95.7Overall Scoreo3
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| o3E2E=true, Param Size=-2025.05 | 95.7 | — | — | |
| BVSModel=Qwen3-VL-8B2026.07 | 93.2 | 95.7 | 89.5 | |
| BVS-Qwen3VL-8BParameters=8B, Backbone=Qwen3-VL2026.07 | 93.2 | 95.7 | 89.5 | |
| BVSModel=InternVL3.5-8B2026.07 | 91.6 | 93 | 89.5 | |
| BVSModel=Qwen3-VL-4B2026.07 | 91.1 | 96.5 | 82.9 | |
| TreeVGR-7BParameters=7B2026.07 | 91.1 | 94 | 87 | |
| ZoomEyeE2E=false, Param Size=7B2025.05 | 90.6 | 93.9 | 85.5 | |
| Simple O3-7BParameters=7B2026.07 | 90.4 | — | — | |
| DeepEyesE2E=true, Param Size=7B2025.05 | 90.1 | 91.3 | 88.2 | |
| ZoomEyeModel=Qwen3-VL-8B2026.07 | 90.1 | 92.2 | 86.8 | |
| DeepEyes-7BParameters=7B2026.07 | 90.1 | 91.3 | 88.2 | |
| Qwen3-VL-8B-w toolParameters=8B, Tool use=true2026.07 | 90.1 | — | — | |
| RAPModel=Qwen3-VL-8B2026.07 | 88.5 | 88.7 | 88.2 | |
| Mini-o3-7BParameters=7B2026.07 | 88.2 | — | — | |
| Qwen2.5-VL*E2E=true, Param Size=32B2025.05 | 87.9 | 87.8 | 88.1 | |
| ZoomEyeModel=Qwen3-VL-4B2026.07 | 87.4 | 91.3 | 81.6 | |
| RAPModel=Qwen3-VL-4B2026.07 | 84.8 | 87.9 | 80.3 | |
| ViCropModel=Qwen3-VL-8B2026.07 | 84.8 | 86.1 | 82.9 | |
| Pixel Reasoner-7BParameters=7B2026.07 | 84.3 | — | — | |
| ZoomEyeModel=InternVL3.5-8B2026.07 | 83.2 | 85.2 | 80.3 | |
| w/o reasoningModel=Qwen3-VL-8B2026.07 | 82.7 | 82.6 | 82.9 | |
| BVSModel=InternVL3.5-4B2026.07 | 82.2 | 83.5 | 80.3 | |
| ViCropModel=Qwen3-VL-4B2026.07 | 82.2 | 85.2 | 77.6 | |
| RAPModel=InternVL3.5-8B2026.07 | 82.2 | 82.6 | 81.6 | |
| Thyme-7BParameters=7B2026.07 | 82.2 | 83.5 | 80.3 | |
| DyFoE2E=false, Param Size=7B2025.05 | 81.2 | 80 | 82.9 | |
| Pixel-ReasonerE2E=true, Param Size=7B2025.05 | 80.6 | 83.5 | 76.3 | |
| ZoomEyeModel=InternVL3.5-4B2026.07 | 79.6 | 80.9 | 77.6 | |
| w/o reasoningModel=Qwen3-VL-4B2026.07 | 78.5 | 80 | 76.3 | |
| ViCropModel=InternVL3.5-8B2026.07 | 78.5 | 80 | 76.3 | |
| RAPModel=InternVL3.5-4B2026.07 | 76.4 | 80 | 71.1 | |
| SEALE2E=false, Param Size=7B2025.05 | 75.4 | 74.8 | 76.3 | |
| LLaVA-OneVisionE2E=true, Param Size=7B2025.05 | 75.4 | 75.7 | 75 | |
| ViCropModel=InternVL3.5-4B2026.07 | 75.4 | 77.4 | 72.4 | |
| Qwen2.5-VL*E2E=true, Param Size=7B2025.05 | 71.2 | 73.9 | 67.1 | |
| w/o reasoningModel=InternVL3.5-8B2026.07 | 68.1 | 65.2 | 72.4 | |
| GPT-4oE2E=true, Param Size=-2025.05 | 66 | — | — | |
| w/o reasoningModel=InternVL3.5-4B2026.07 | 62.3 | 67 | 55.3 |