Large Multi-modal Model Evaluation on MME
1,514.46Perception ScoreREVIS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| REVISModel=LLaVA-NEXT2026.02 | 1,514.46 | 302.86 | 1,817.32 | |
| RegularModel=LLaVA-NEXT, Decoding=Standard greedy decoding2026.02 | 1,512.19 | 291.79 | 1,803.98 | |
| M3IDModel=LLaVA-NEXT2026.02 | 1,512.19 | 291.79 | 1,803.98 | |
| AGLAModel=LLaVA-NEXT2026.02 | 1,511.94 | 291.79 | 1,803.73 | |
| OnlyModel=LLaVA-NEXT2026.02 | 1,510.88 | 298.57 | 1,809.45 | |
| AGLAModel=LLaVA-1.52026.02 | 1,486.37 | 287.14 | 1,773.51 | |
| REVISModel=LLaVA-1.52026.02 | 1,481.56 | 305.71 | 1,787.27 | |
| VCDModel=LLaVA-1.52026.02 | 1,472.91 | 298.93 | 1,771.84 | |
| VCDModel=LLaVA-NEXT2026.02 | 1,463.05 | 291.07 | 1,754.12 | |
| SVCBase Model=LLaVA-1.52026.03 | 1,456.28 | 332.86 | — | |
| RegularModel=LLaVA-1.5, Decoding=Standard greedy decoding2026.02 | 1,455.42 | 298.93 | 1,754.33 | |
| M3IDModel=LLaVA-1.52026.02 | 1,455.42 | 298.93 | 1,754.33 | |
| VISTABase Model=LLaVA-1.52026.03 | 1,450.54 | 329.08 | — | |
| VCDBase Model=LLaVA-1.52026.03 | 1,445.13 | 324.25 | — | |
| VanillaBase Model=LLaVA-1.52026.03 | 1,444.23 | 323.14 | — | |
| OnlyModel=LLaVA-1.52026.02 | 1,424.71 | 281.07 | 1,705.78 | |
| VTIModel=LLaVA-NEXT2026.02 | 1,416.51 | 268.21 | 1,684.72 | |
| VTIModel=LLaVA-1.52026.02 | 1,240.31 | 267.14 | 1,507.45 | |
| SVCBase Model=InstructBlip2026.03 | 1,192.76 | 240.83 | — | |
| VISTABase Model=InstructBlip2026.03 | 1,189.23 | 236.25 | — | |
| VCDBase Model=InstructBlip2026.03 | 1,182.34 | 234.95 | — | |
| VanillaBase Model=InstructBlip2026.03 | 1,181.15 | 233.71 | — | |
| ID-Selection§Token Budget=320 (88.9%)2026.04 | — | — | 1,817 | |
| ID-Selection§Token Budget=160 (94.4%)2026.04 | — | — | 1,756 | |
| LLaVA-Next-7BToken Budget=2880 (100%)2026.04 | — | — | 1,851 |