Multi-modal Understanding on MMBench
87.1AccuracyBAGEL (Ours)
Evaluation results
| Method | Links | |
|---|---|---|
| BAGEL (Ours)Baseline=BAGEL, Params=7B active2026.06 | 87.1 | |
| BLIP3o-8B (Ours)Baseline=BLIP3o-8B, Params=8B2026.06 | 86.1 | |
| BAGELBaseline=BAGEL, Params=7B active2026.06 | 85 | |
| UniCornBaseline=BAGEL, Params=7B active2026.06 | 84.1 | |
| VARGPT-v1.1 (Ours)Baseline=VARGPT-v1.1, Params=7B+2B2026.06 | 83.7 | |
| BLIP3o-8BBaseline=BLIP3o-8B, Params=8B2026.06 | 83.5 | |
| UniGameBaseline=Janus-Pro-7B, Params=7B2026.06 | 83.2 | |
| InternVL2.5-Pt + IRABackbone=InternVL2.5, Training stage=IRA2026.07 | 81.8 | |
| VARGPT-v1.1Baseline=VARGPT-v1.1, Params=7B+2B2026.06 | 81 | |
| InternVL2-Pt + IRABackbone=InternVL2, Training stage=IRA2026.07 | 81 | |
| InternVL2.5-Pt + SFTBackbone=InternVL2.5, Training stage=SFT2026.07 | 80.6 | |
| InternVL2-Pt + SFTBackbone=InternVL2, Training stage=SFT2026.07 | 80.4 | |
| SUDERBaseline=Janus-Pro-7B, Params=7B2026.06 | 80.1 | |
| LLaVA-OV-Pt + IRABackbone=LLaVA-OV, Training stage=IRA2026.07 | 79.6 | |
| LLaVA-OV-Pt + SFTBackbone=LLaVA-OV, Training stage=SFT2026.07 | 79.5 | |
| Janus-Pro-7BBaseline=Janus-Pro-7B, Params=7B2026.06 | 79.2 | |
| TokenFlowBaseline=TokenFlow2026.06 | 76.8 | |
| MetaMorphBaseline=MetaMorph2026.06 | 75.2 | |
| ILLUMEBaseline=Vicuna-7B, Params=7B2026.06 | 75.1 | |
| InternVL2.5-PtBackbone=InternVL2.5, Training stage=Pre-trained2026.07 | 72.7 | |
| InternVL2-PtBackbone=InternVL2, Training stage=Pre-trained2026.07 | 70.9 | |
| SEED-XBaseline=SEED-X2026.06 | 70.1 | |
| JanusBaseline=Janus2026.06 | 69.4 | |
| LLaVA-OV-PtBackbone=LLaVA-OV, Training stage=Pre-trained2026.07 | 67.3 | |
| VanillaToken Budget=576, Base Model=LLaVA-1.5-7B2026.06 | 64.7 | |
| CLSEToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 63.6 | |
| PDropToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 63.2 | |
| CLSEToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 63 | |
| HiREDToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 62.8 | |
| SparseVLMToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 62.5 | |
| FiCoCo-VToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 62.3 | |
| CLSEToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 61.9 | |
| HiREDToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 61.5 | |
| FastVToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 61.2 | |
| PDropToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 61.1 | |
| FiCoCo-VToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 61.1 | |
| ToMeToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 60.5 | |
| FiCoCo-VToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 60.3 | |
| HiREDToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 60.2 | |
| SparseVLMToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 60 | |
| PruMergeToken Budget=192, Base Model=LLaVA-1.5-7B2026.06 | 59.6 | |
| Emu3Baseline=Emu32026.06 | 58.5 | |
| PruMergeToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 58.1 | |
| SparseVLMToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 56.2 | |
| FastVToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 56.1 | |
| PruMergeToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 55.3 | |
| ToMeToken Budget=128, Base Model=LLaVA-1.5-7B2026.06 | 53.3 | |
| FastVToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 48 | |
| ToMeToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 43.7 | |
| PDropToken Budget=64, Base Model=LLaVA-1.5-7B2026.06 | 33.3 | |
| ChameleonBaseline=Chameleon2026.06 | 19.8 |