Visual Perception and Reasoning on V* Bench
98.3Attribute ScoreHuman
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Human2026.04 | 98.3 | 100 | 99 | |
| TTSPBackbone=Qwen3-VL-8B-Instruct2026.04 | 95.7 | 93.4 | 94.8 | |
| TTSPBackbone=Qwen3-VL-4B-Instruct2026.04 | 94.8 | 92.1 | 93.7 | |
| RTWIBackbone=Qwen3-VL-8B-Instruct2026.04 | 94.8 | 92.1 | 93.7 | |
| RTWIBackbone=Qwen3-VL-4B-Instruct2026.04 | 93.9 | 90.8 | 92.7 | |
| CISCBackbone=Qwen3-VL-8B-Instruct2026.04 | 93.9 | 92.1 | 93.2 | |
| DeepconfBackbone=Qwen3-VL-8B-Instruct2026.04 | 93.9 | 92.1 | 93.2 | |
| CISCBackbone=Qwen3-VL-4B-Instruct2026.04 | 93 | 90.8 | 92.1 | |
| DeepconfBackbone=Qwen3-VL-4B-Instruct2026.04 | 93 | 92.1 | 92.6 | |
| SCBackbone=Qwen3-VL-8B-Instruct2026.04 | 93 | 90.8 | 92.1 | |
| Self-Cer.Backbone=Qwen3-VL-8B-Instruct2026.04 | 93 | 89.5 | 91.6 | |
| SCBackbone=Qwen3-VL-4B-Instruct2026.04 | 92.2 | 90.8 | 91.6 | |
| TRTBackbone=Qwen3-VL-4B-Instruct2026.04 | 92.2 | 90.8 | 91.6 | |
| BaseBackbone=Qwen3-VL-8B-Instruct2026.04 | 92.2 | 89.5 | 91.1 | |
| TRTBackbone=Qwen3-VL-8B-Instruct2026.04 | 92.2 | 90.8 | 91.6 | |
| Self-Ref.Backbone=Qwen3-VL-8B-Instruct2026.04 | 92.2 | 90.8 | 91.6 | |
| ZoomEyeBackbone=Qwen3-VL-4B-Instruct2026.03 | 92.17 | 86.84 | 90.05 | |
| DeepEyes2026.04 | 91.3 | 88.2 | 90.1 | |
| BaseBackbone=Qwen3-VL-4B-Instruct2026.04 | 91.3 | 89.5 | 90.6 | |
| Self-Ref.Backbone=Qwen3-VL-4B-Instruct2026.04 | 91.3 | 89.5 | 90.6 | |
| Self-Cer.Backbone=Qwen3-VL-4B-Instruct2026.04 | 91.3 | 89.5 | 90.6 | |
| DDP2026.04 | 89.2 | 89.5 | 89.3 | |
| SIEVEBackbone=Qwen3-VL-8B-Instruct2026.03 | 88.7 | 86.84 | 87.96 | |
| GRPOBackbone=Qwen3-VL-8B-Instruct2026.03 | 86.96 | 86.84 | 86.91 | |
| DyFoBackbone=Qwen3-VL-4B-Instruct2026.03 | 86.09 | 75 | 81.68 | |
| Zoom-RefineBackbone=Qwen3-VL-4B-Instruct2026.03 | 86.09 | 73.47 | 81.17 | |
| DyFoBackbone=Qwen3-VL-8B-Instruct2026.03 | 86.09 | 78.95 | 83.25 | |
| SCOLARtraining_stage=SFT only2026.05 | 86.09 | 81.58 | 84.29 | |
| SCOLAR-7BBackbone=Qwen2.5-VL-7B2026.05 | 86.09 | 80.26 | 83.77 | |
| Vanilla*Backbone=Qwen3-VL-8B-Instruct2026.03 | 84.35 | 80.26 | 82.72 | |
| Deepeyes2026.05 | 84.35 | 81.58 | 83.25 | |
| Thyme2026.04 | 83.5 | 80.3 | 82.2 | |
| Pixel-Reasoner2026.04 | 83.5 | 76.3 | 80.6 | |
| Zoom-RefineBackbone=Qwen3-VL-8B-Instruct2026.03 | 83.48 | 82.89 | 83.25 | |
| Monet2026.05 | 83.48 | 82.89 | 83.25 | |
| GRPOBackbone=Qwen3-VL-4B-Instruct2026.03 | 81.74 | 90.79 | 85.34 | |
| SIEVEBackbone=Qwen3-VL-4B-Instruct2026.03 | 81.74 | 92.11 | 85.86 | |
| ZoomEyeBackbone=Qwen3-VL-8B-Instruct2026.03 | 81.74 | 73.68 | 78.35 | |
| LVR2026.05 | 81.7 | 79 | 80.6 | |
| Qwen2.5-VL-7B2026.05 | 80 | 76.31 | 78.53 | |
| Vanilla*Backbone=Qwen3-VL-4B-Instruct2026.03 | 74.78 | 82.89 | 78.01 | |
| GPT-4o2026.04 | 72.2 | 60.5 | 67.5 | |
| GPT-4o2026.05 | 72.2 | 60.5 | 67.5 | |
| GPT-4V2026.05 | 61.74 | 67.11 | 63.87 | |
| GPT-4V2026.04 | 51.3 | 60.5 | 55 | |
| LLaVA-1.5-7BParameters=7B2026.04 | 43.5 | 56.6 | 48.7 | |
| Gemini-1.5-Pro2026.04 | 40.9 | 59.2 | 48.2 | |
| Bard2026.04 | 31.3 | 46.1 | 37.2 | |
| InstructBLIP2026.04 | 25.2 | 47.4 | 34 |