Multimodal Perception on MME
2,242.7Perception ScoreInternLM-XComposer2
Evaluation results
| Method | Links | ||
|---|---|---|---|
| InternLM-XComposer2Number of Parameters=7B2024.01 | 2,242.7 | — | |
| Qwen-VL-Plus2024.01 | 2,183.3 | — | |
| WeMMNumber of Parameters=6B2024.01 | 2,066.6 | — | |
| Gemini-Pro2024.01 | 1,933.3 | — | |
| GPT-4V2024.01 | 1,926.5 | — | |
| ICLABackbone=Qwen3-VL-8B2026.02 | 1,740 | — | |
| DeCoBackbone=Qwen3-VL-8B2026.02 | 1,734 | — | |
| VanillaBackbone=Qwen3-VL-8B2026.02 | 1,726 | — | |
| VCDBackbone=Qwen3-VL-8B2026.02 | 1,726 | — | |
| VDDBackbone=Qwen3-VL-8B2026.02 | 1,720 | — | |
| DoLABackbone=Qwen3-VL-8B2026.02 | 1,705 | — | |
| DAMOBackbone=Qwen3-VL-8B2026.02 | 1,704 | — | |
| SugarLLM=Vicuna-7B, Resolution=3362024.11 | 1,550.8 | — | |
| TokenFlow-XLType=Und. & Gen., # Params=13B2025.09 | 1,545.9 | — | |
| VILA-7BLLM=Llama-2-7B, Resolution=3362024.11 | 1,533 | — | |
| LLaVA-1.5LLM=Vicuna-7B, Resolution=3362024.11 | 1,510.7 | — | |
| LLaVA-v1.5Type=Und. Only, # Params=7B2025.09 | 1,510.7 | — | |
| OursType=Und. & Gen., # Params=1.5B2025.09 | 1,492.7 | — | |
| Qwen-VL-ChatLLM=Qwen-7B, Resolution=4482024.11 | 1,487.5 | — | |
| Qwen-VL-ChatType=Und. Only, # Params=7B2025.09 | 1,487.5 | — | |
| FUDOKIType=Und. & Gen., # Params=1.5B2025.09 | 1,485.4 | — | |
| LLaVA-1.5-7BVision Encoder=MACCO-CLIP2026.06 | 1,452.3 | — | |
| LLaVA-1.5-7BVision Encoder=vanilla CLIP2026.06 | 1,447.1 | — | |
| ILLUMEType=Und. & Gen., # Params=7B2025.09 | 1,445.3 | — | |
| Janus-Pro-1BType=Und. & Gen., # Params=1.5B2025.09 | 1,444 | — | |
| MobileVLM-V2Type=Und. Only, # Params=2.7B2025.09 | 1,440.5 | — | |
| VILA-UType=Und. & Gen., # Params=7B2025.09 | 1,401.8 | — | |
| JanusType=Und. & Gen., # Params=1.5B2025.09 | 1,338 | — | |
| LLaVA-PhiType=Und. Only, # Params=2.7B2025.09 | 1,335.1 | — | |
| MobileVLM-V2Type=Und. Only, # Params=1.4B2025.09 | 1,302.8 | — | |
| BLIP-2LLM=Vicuna-13B, Resolution=2242024.11 | 1,293.8 | — | |
| MobileVLMType=Und. Only, # Params=2.7B2025.09 | 1,288.9 | — | |
| Emu3-ChatType=Und. Only, # Params=8B2025.09 | 1,244 | — | |
| InstructBLIPLLM=Vicuna-13B, Resolution=2242024.11 | 1,212.8 | — | |
| InstructBLIPType=Und. Only, # Params=13B2025.09 | 1,212.8 | — | |
| MobileVLMType=Und. Only, # Params=1.4B2025.09 | 1,196.2 | — | |
| LLaVA-Phi-1.5Type=Und. Only, # Params=1.3B2025.09 | 1,128 | — | |
| D-DitType=Und. & Gen., # Params=2.0B2025.09 | 1,124.7 | — | |
| Show-o-512Type=Und. & Gen., # Params=1.3B2025.09 | 1,097.2 | — | |
| Show-o-256Type=Und. & Gen., # Params=1.3B2025.09 | 948.4 | — | |
| LLaVAType=Und. Only, # Params=7B2025.09 | 809.6 | — | |
| LoRA-onlyTrainable Params=7.6M2026.03 | 83.2 | — | |
| IADA +LoRATrainable Params=7.7M2026.03 | 83.2 | — | |
| AttnRes+LoRATrainable Params=7.7M2026.03 | 82.7 | — | |
| Pre-trainedTrainable Params=02026.03 | 82.3 | — | |
| ClearSightBase Model=LLaVA-1.52026.01 | — | 1,569 | |
| ClearSightBase Model=LLaVA-NeXT2026.01 | — | 1,602 | |
| ClearSightBase Model=Qwen2-VL2026.01 | — | 2,001 | |
| CRoPSBase Model=LLaVA-1.52026.01 | — | 1,662 | |
| CRoPSBase Model=LLaVA-NeXT2026.01 | — | 1,779 | |
| CRoPSBase Model=Qwen2-VL2026.01 | — | 2,184 | |
| ICDBase Model=LLaVA-1.52026.01 | — | 1,605 | |
| ICDBase Model=LLaVA-NeXT2026.01 | — | 1,664 | |
| ICDBase Model=Qwen2-VL2026.01 | — | 2,060 | |
| M3IDBase Model=LLaVA-1.52026.01 | — | 1,607 | |
| M3IDBase Model=LLaVA-NeXT2026.01 | — | 1,683 | |
| M3IDBase Model=Qwen2-VL2026.01 | — | 2,090 | |
| OPERABase Model=LLaVA-1.52026.01 | — | 1,615 | |
| OPERABase Model=LLaVA-NeXT2026.01 | — | 1,650 | |
| OPERABase Model=Qwen2-VL2026.01 | — | 2,080 | |
| SamplingBase Model=LLaVA-1.52026.01 | — | 1,601 | |
| SamplingBase Model=LLaVA-NeXT2026.01 | — | 1,669 | |
| SamplingBase Model=Qwen2-VL2026.01 | — | 2,058 | |
| SIDBase Model=LLaVA-1.52026.01 | — | 1,634 | |
| SIDBase Model=LLaVA-NeXT2026.01 | — | 1,607 | |
| SIDBase Model=Qwen2-VL2026.01 | — | 2,097 | |
| VCDBase Model=LLaVA-1.52026.01 | — | 1,622 | |
| VCDBase Model=LLaVA-NeXT2026.01 | — | 1,630 | |
| VCDBase Model=Qwen2-VL2026.01 | — | 2,070 |