Multimodal Understanding on MMBench v1.1 (dev)
91.2MMBench ScoreUniDFlow
Evaluation results
| Method | Links | ||
|---|---|---|---|
| UniDFlowParams=4B2026.02 | 91.2 | — | |
| MammothModa-2Params=4B2026.02 | 86.6 | — | |
| EMMAParams=4B2026.02 | 85.8 | — | |
| Qwen3-VLParams=4B2026.02 | 85.1 | — | |
| BAGELParams=7B2026.02 | 85 | — | |
| MudditParams=4B2026.02 | 82.8 | — | |
| Janus-ProParams=7B2026.02 | 79.2 | — | |
| Qwen2.5-VLParams=3B2026.02 | 79.1 | — | |
| BLIP-3Params=4B2026.02 | 76.8 | — | |
| TokenFlow-XLParams=13B2026.02 | 68.9 | — | |
| LLaVABackbone=LLaVA-NeXT-7B, Token Budget=2880, Retention Ratio=100%2025.08 | 67.4 | — | |
| HiPrune++Backbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 67.2 | — | |
| HiPruneBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 67 | — | |
| TRIMBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 66.8 | — | |
| VILA-UParams=7B2026.02 | 66.6 | — | |
| VisionZipBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 66.3 | — | |
| HiPrune++Backbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 66.2 | — | |
| HiREDBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 66 | — | |
| HiPruneBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 65.3 | — | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 65.2 | — | |
| DivPruneBackbone=LLaVA-NeXT-7B, Token Budget=640, Retention Ratio=22.2%2025.08 | 65 | — | |
| HiREDBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 64.2 | — | |
| DivPruneBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 63.9 | — | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 63.8 | — | |
| TRIMBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 63.5 | — | |
| VisionZipBackbone=LLaVA-NeXT-7B, Token Budget=320, Retention Ratio=11.1%2025.08 | 63.1 | — | |
| DivPruneBackbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 62.5 | — | |
| TRIMBackbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 61.6 | — | |
| HiPrune++Backbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 61.5 | — | |
| VisionZipBackbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 60.1 | — | |
| HiPruneBackbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 59.8 | — | |
| VisPrunerBackbone=LLaVA-NeXT-7B, Token Budget=160, Retention Ratio=5.6%2025.08 | 59.2 | — | |
| OmniGen-v2Params=8B2026.02 | 53.1 | — | |
| DeepSeek-VL2Params=4B2026.02 | 51.1 | — | |
| ChameleonParams=7B2026.02 | 35.7 | — | |
| LLaVA-1.6-Mistral-7BBackbone=Mistral-7B2026.02 | — | 68.18 | |
| LLaVA-1.6-Vicuna-13BBackbone=Vicuna-13B2026.02 | — | 69.8 | |
| LLaVA-Bielik-11b-v2.6Parameters=11B, Version=v2.6, Training=Parameter-efficient LoRA adaptation2026.02 | — | 78.24 | |
| LLaVA-PLLuM-12b-ncParameters=12B, Training=Parameter-efficient LoRA adaptation2026.02 | — | 79.35 | |
| LLaVA-PLLuM-12b-nc-250715Parameters=12B, Training=Parameter-efficient LoRA adaptation2026.02 | — | 76.73 | |
| PaliGemma2-10BParameters=10B2026.02 | — | 78.39 | |
| Pixtral-12BParameters=12B, Training=Full fine-tuning2026.02 | — | 82.06 | |
| Qwen2.5-VL-7BParameters=7B2026.02 | — | 75.56 |