Comprehensive Multi-modal Evaluation on MME
2,306.5Total ScoreQwen2.5-VL-7B
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Qwen2.5-VL-7BToken Retention Rate=100%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,306.5 | — | — | — | — | — | — | — | |
| ERAToken Retention Rate=50%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,305.5 | — | — | — | — | — | — | — | |
| VisionZipToken Retention Rate=50%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,292.6 | — | — | — | — | — | — | — | |
| FastVToken Retention Rate=50%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,286.7 | — | — | — | — | — | — | — | |
| FastVToken Retention Rate=20%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,225.4 | — | — | — | — | — | — | — | |
| ERAToken Retention Rate=20%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,224.3 | — | — | — | — | — | — | — | |
| CDPrunerToken Retention Rate=50%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,221.6 | — | — | — | — | — | — | — | |
| PTPModel=InternVL2-8B, Pruning Method=PTP, Pruning Ratio (r)=0.5, Balancing Weight (alpha)=0.52025.09 | 2,220.5 | — | — | — | — | — | — | — | |
| VisionZipToken Retention Rate=20%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,212.2 | — | — | — | — | — | — | — | |
| InternVL2-8BModel=InternVL2-8B, Pruning Method=None2025.09 | 2,205.3 | — | — | — | — | — | — | — | |
| CDPrunerToken Retention Rate=20%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,068.2 | — | — | — | — | — | — | — | |
| ERAToken Retention Rate=10%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,041.2 | — | — | — | — | — | — | — | |
| VisionZipToken Retention Rate=10%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 2,019.9 | — | — | — | — | — | — | — | |
| FastVToken Retention Rate=10%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 1,975.4 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,908 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,908 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=960, Pruning Ratio=66.7%2026.02 | 1,902 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=960, Pruning Ratio=66.7%2026.02 | 1,902 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Retained Tokens=960, Pruning Ratio=66.7%2026.02 | 1,885 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Tokens Retained=960, Pruning Ratio=66.7%2026.02 | 1,885 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,882 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,882 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=960, Pruning Ratio=66.7%2026.02 | 1,880 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=960, Pruning Ratio=66.7%2026.02 | 1,880 | — | — | — | — | — | — | — | |
| PTPModel=InternVL2-2B, Pruning Method=PTP, Pruning Ratio (r)=0.5, Balancing Weight (alpha)=0.52025.09 | 1,879.7 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,876 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,876 | — | — | — | — | — | — | — | |
| CDPrunerToken Retention Rate=10%, Backbone=Qwen2.5-VL-7B, Resolution Strategy=Dynamic Resolution2026.06 | 1,872.3 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,871 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,871 | — | — | — | — | — | — | — | |
| LLaVA-1.5-7BRetained visual tokens=576, Pruning ratio=0%2026.03 | 1,862 | — | — | — | — | — | — | 100 | |
| DARTRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,856 | — | — | — | — | — | — | 98.6 | |
| DARTRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,845 | — | — | — | — | — | — | 97 | |
| VisionZipBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,844 | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,844 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-NeXT-13B, Retained Tokens=960, Pruning Ratio=66.7%2026.02 | 1,840 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-NeXT-13B, Tokens Retained=960, Pruning Ratio=66.7%2026.02 | 1,840 | — | — | — | — | — | — | — | |
| LLaVA-NeXT-13BBackbone=LLaVA-NeXT-13B, Retained Tokens=2880, Pruning Ratio=0%2026.02 | 1,837 | — | — | — | — | — | — | — | |
| LLaVA-NeXT-13BBackbone=LLaVA-NeXT-13B, Tokens Retained=2880, Pruning Ratio=0%2026.02 | 1,837 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,831 | — | — | — | — | — | — | — | |
| VisPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,831 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,830 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,830 | — | — | — | — | — | — | — | |
| LLaVA-1.5-13BBackbone=LLaVA-1.5-13B, Retained Tokens=576, Pruning Ratio=0%2026.02 | 1,828 | — | — | — | — | — | — | — | |
| InternVL2-2BModel=InternVL2-2B, Pruning Method=None2025.09 | 1,821.7 | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,816 | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,816 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-1.5-13B, Retained Tokens=192, Pruning Ratio=66.7%2026.02 | 1,805 | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-NeXT-13B, Retained Tokens=640, Pruning Ratio=77.8%2026.02 | 1,804 | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-NeXT-13B, Tokens Retained=640, Pruning Ratio=77.8%2026.02 | 1,804 | — | — | — | — | — | — | — | |
| SCOPERetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,804 | — | — | — | — | — | — | 98.9 | |
| ResPruneRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,795 | — | — | — | — | — | — | 99.4 | |
| ResPruneRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,792 | — | — | — | — | — | — | 99.3 | |
| HoloVBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,789 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,789 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,789 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,789 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,778 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=192, Pruning Ratio=66.7%2026.02 | 1,776 | — | — | — | — | — | — | — | |
| SCOPERetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,776 | — | — | — | — | — | — | 98.3 | |
| VisionZipBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,770 | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,770 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,768 | — | — | — | — | — | — | — | |
| VisionZIPRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,767 | — | — | — | — | — | — | 98.2 | |
| PDropRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,766 | — | — | — | — | — | — | 96.7 | |
| DARTRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,765 | — | — | — | — | — | — | 94.3 | |
| VisPrunerRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,762 | — | — | — | — | — | — | 97.9 | |
| VisionZIPRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,762 | — | — | — | — | — | — | 97.1 | |
| AdaPruneRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,755 | — | — | — | — | — | — | 98.3 | |
| HoloVBackbone=LLaVA-1.5-13B, Retained Tokens=192, Pruning Ratio=66.7%2026.02 | 1,754 | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-NeXT-13B, Retained Tokens=320, Pruning Ratio=88.9%2026.02 | 1,753 | — | — | — | — | — | — | — | |
| DivPruneBackbone=LLaVA-NeXT-13B, Tokens Retained=320, Pruning Ratio=88.9%2026.02 | 1,753 | — | — | — | — | — | — | — | |
| DivPruneRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,751 | — | — | — | — | — | — | 97.9 | |
| VisPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=192, Pruning Ratio=66.7%2026.02 | 1,750 | — | — | — | — | — | — | — | |
| FSRBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,750 | — | — | — | — | — | — | — | |
| CDPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,744 | — | — | — | — | — | — | — | |
| TRIMRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,743 | — | — | — | — | — | — | 96.5 | |
| ResPruneRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,740 | — | — | — | — | — | — | 98 | |
| HiREDRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,737 | — | — | — | — | — | — | 94.6 | |
| VisPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,736 | — | — | — | — | — | — | — | |
| HoloVBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,731 | — | — | — | — | — | — | — | |
| FastVBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,722 | — | — | — | — | — | — | — | |
| SparseVLMRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,721 | — | — | — | — | — | — | 96.6 | |
| DivPruneBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,720 | — | — | — | — | — | — | — | |
| AdaPruneRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,715 | — | — | — | — | — | — | 97.1 | |
| DivPruneBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,713 | — | — | — | — | — | — | — | |
| VisionZipBackbone=LLaVA-1.5-13B, Retained Tokens=128, Pruning Ratio=77.8%2026.02 | 1,710 | — | — | — | — | — | — | — | |
| HiREDRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,710 | — | — | — | — | — | — | 93.2 | |
| SCOPERetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,698 | — | — | — | — | — | — | 96.8 | |
| SparseVLMRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,696 | — | — | — | — | — | — | 94.5 | |
| VisPrunerBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,694 | — | — | — | — | — | — | — | |
| DivPruneRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,694 | — | — | — | — | — | — | 97.2 | |
| VisionZIPRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,690 | — | — | — | — | — | — | 94.4 | |
| HoloVBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,683 | — | — | — | — | — | — | — | |
| TRIMRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,680 | — | — | — | — | — | — | 94.4 | |
| VisPrunerRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,674 | — | — | — | — | — | — | 95.3 | |
| DivPruneRetained visual tokens=64, Pruning ratio=88.9%2026.03 | 1,657 | — | — | — | — | — | — | 95.7 | |
| PDropRetained visual tokens=128, Pruning ratio=77.8%2026.03 | 1,644 | — | — | — | — | — | — | 94.6 | |
| PruMergeRetained visual tokens=192, Pruning ratio=66.7%2026.03 | 1,632 | — | — | — | — | — | — | 91.5 | |
| VisionZipBackbone=LLaVA-1.5-13B, Retained Tokens=64, Pruning Ratio=88.9%2026.02 | 1,628 | — | — | — | — | — | — | — |