Multi-modal Reasoning on MMVet
51.5ScoreQwen2-VL-2B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2-VL-2B#Params=2B, TFLOPs=0.42025.06 | 51.5 | |
| LaTtE-Flow#Params=2B, TFLOPs=0.42025.06 | 51.5 | |
| AsyMoE-LLaMA3LLM Backbone=LLaMA-3-8B, Act Param=10.8B2026.05 | 49.2 | |
| CuMoLLM Backbone=Mixtral-8×7B, Act Param=13.5B2026.05 | 48.7 | |
| EMU2 Chat#Params=34B, TFLOPs=5.42025.06 | 48.5 | |
| MoIIE-LLaMA3LLM Backbone=LLaMA-3-8B, Act Param=11.3B2026.05 | 47.3 | |
| MM1LLM Backbone=MM1-7B-MoE2026.05 | 47 | |
| Mini-GeminiLLM Backbone=Mixtral-8×7B, Act Param=13.5B2026.05 | 45.8 | |
| AsyMoE-Phi3LLM Backbone=Phi-3-mini, Act Param=4.1B2026.05 | 45.8 | |
| LLaVA-NeXTLLM Backbone=Vicuna-7B, Act Param=7.1B2026.05 | 43.9 | |
| ContextMoE-Phi2LLM Backbone=Phi-2, Act Param=4.1B2026.05 | 43.5 | |
| Seed-X#Params=17B, TFLOPs=11.12025.06 | 43 | |
| MoIIE-Phi3LLM Backbone=Phi-3-mini, Act Param=5.5B2026.05 | 42.8 | |
| Deepseek-VLLLM Backbone=Deepseek-7B2026.05 | 41.5 | |
| SPHINX-XLLM Backbone=Mixtral-8×7B2026.05 | 40.9 | |
| Mini-GeminiLLM Backbone=Vicuna-7B, Act Param=7.3B2026.05 | 40.8 | |
| CDPrunerRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 40.7 | |
| EADPRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 40.1 | |
| LLaVA-NeXT-7BRetained Tokens=2,880, Token Reduction Ratio=100%2026.07 | 40 | |
| Janus Pro#Params=1.5B, TFLOPs=0.82025.06 | 39.8 | |
| EADPRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 39.4 | |
| VisionZipRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 39.2 | |
| MousiLLM Backbone=Vicuna-7B, Act Param=7.9B2026.05 | 38.4 | |
| HiPruneRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 38.4 | |
| DivPruneRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 38 | |
| CDPrunerRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 37.9 | |
| ShareGPT-4VLLM Backbone=Vicuna-7B2026.05 | 37.6 | |
| DARTRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 37.3 | |
| EMU3#Params=8B, TFLOPs=4.12025.06 | 37.2 | |
| FastVRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 37.1 | |
| TRIMRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 36.9 | |
| DARTRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 36.3 | |
| mPLUG-Owl2LLM Backbone=LLaMA-2-7B2026.05 | 36.2 | |
| MoE-LLaVA-Phi2LLM Backbone=Phi-2, Act Param=3.6B2026.05 | 35.9 | |
| SparseVLMRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 35.7 | |
| EADPRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 35.6 | |
| PDropRetained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 35.4 | |
| VisionZipRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 35.2 | |
| CDPrunerRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 35 | |
| DivPruneRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 34.8 | |
| HiPruneRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 34.7 | |
| Janus#Params=1.5B, TFLOPs=0.82025.06 | 34.3 | |
| PruMerge+Retained Tokens=640, Token Reduction Ratio=↓ 77.8%2026.07 | 33.6 | |
| VILA-U#Params=7B, TFLOPs=3.62025.06 | 33.5 | |
| TRIMRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 33.5 | |
| AnchorPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 32.8 | |
| VisionZipBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 32.7 | |
| DivPruneRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 32.4 | |
| PruMerge+Retained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 32.3 | |
| DARTRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 32 | |
| SparseVLMRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 31.9 | |
| VisionZipRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 31.6 | |
| LLaVA-1.5-7BBackbone=LLaVA-1.5-7B, Visual Token Budget=5762026.07 | 31.3 | |
| LLaVA-1.5-7BLLM Backbone=Vicuna-7B, Act Param=7.1B2026.05 | 30.5 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 30.4 | |
| TRIMRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 30.1 | |
| AnchorPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 30.1 | |
| VisionZipBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 30 | |
| HiPruneRetained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 29.2 | |
| DivPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 29.2 | |
| AnchorPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=322026.07 | 28.7 | |
| FastVBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 28.2 | |
| CDPrunerBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 28.1 | |
| CDPrunerBackbone=LLaVA-1.5-7B, Visual Token Budget=322026.07 | 27.7 | |
| PruMerge+Retained Tokens=160, Token Reduction Ratio=↓ 94.4%2026.07 | 27.6 | |
| PruMerge+Backbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 26.6 | |
| FastVBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 26.4 | |
| PyramidDropBackbone=LLaVA-1.5-7B, Visual Token Budget=1282026.07 | 26.3 | |
| InstructBLIP-7BLLM Backbone=Vicuna-7B, Act Param=7.9B2026.05 | 26.2 | |
| CDPrunerBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 26.1 | |
| DivPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=322026.07 | 24.3 | |
| DivPruneBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 24.2 | |
| PruMerge+Backbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 24.1 | |
| VisionZipBackbone=LLaVA-1.5-7B, Visual Token Budget=322026.07 | 24 | |
| PDropRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 23.7 | |
| SparseVLMBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 23.6 | |
| PruMerge+Backbone=LLaVA-1.5-7B, Visual Token Budget=322026.07 | 22.1 | |
| FastVRetained Tokens=320, Token Reduction Ratio=↓ 88.9%2026.07 | 21.1 | |
| PyramidDropBackbone=LLaVA-1.5-7B, Visual Token Budget=642026.07 | 20.2 | |
| Chameleon#Params=34B, TFLOPs=17.42025.06 | 9.7 | |
| Chameleon#Params=7B, TFLOPs=3.62025.06 | 8.3 |