Multimodal Understanding on CQA, DVQA, TVQA, IVQA, OCRB, MMB, MMMU, MVista, AI2D, MMS, MMV (test)
85.8MMBEMMA
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| EMMAParams=4B, Mixture-of-experts architecture=true2025.12 | 85.8 | 88 | 95.9 | 79.5 | 77.7 | 89 | 62.5 | 75.8 | 83.9 | 64.8 | 73 | 79.6 | |
| EMMA (W/O)Params=4B, Mixture-of-experts architecture=false2025.12 | 85.6 | 87.8 | 95.5 | 79 | 77.4 | 88.4 | 62 | 75.4 | 83.6 | 64.2 | 72.5 | 79.2 | |
| BAGELParams=7B2025.12 | 85 | — | — | — | — | — | 55.3 | 73.1 | — | — | 67.2 | — | |
| LLaVA-OV-1.5Params=4B2025.12 | 84.2 | 87.1 | 94.4 | — | 76.1 | 80 | 52.7 | 67.9 | 83.6 | 64.9 | — | — | |
| Qwen3-VLParams=4B, Evaluation tool=VLMEvalKit, Checkpoint=instruct-version2025.12 | 83.6 | 82.8 | 95.2 | 80.9 | 79.9 | 88.1 | 67.4 | 73.7 | 83.9 | 69.8 | 66.2 | 79.2 | |
| UniWolrd-V1Params=7B2025.12 | 83.5 | — | — | — | — | — | 58.6 | — | — | — | 67.1 | — | |
| BLIP3-oParams=7B2025.12 | 83.5 | — | — | — | — | — | 58.6 | — | — | — | 66.6 | — | |
| InternVl-3.5Params=4B, Evaluation tool=VLMEvalKit, Checkpoint=instruct-version2025.12 | 81.4 | 86.3 | 91.9 | 77.8 | 77.5 | 81.7 | 60 | 68.4 | 82 | 64 | 76.1 | 77 | |
| Show-o2Params=7B2025.12 | 79.3 | — | — | — | — | — | 48.9 | — | 78.6 | 56.6 | — | — | |
| Janus ProParams=7B2025.12 | 79.2 | — | — | — | — | — | 41 | — | — | — | 50 | — | |
| OminiGen2Params=3B2025.12 | 79.1 | — | — | — | — | — | 53.1 | — | — | — | 61.8 | — | |
| ILLUMEParams=7B2025.12 | 75.1 | 66.7 | 76 | — | 45.5 | 66.9 | 38.2 | — | 71.4 | — | 37 | — | |
| MogaoParams=7B2025.12 | 75 | — | — | — | — | — | 44.2 | — | — | — | — | — | |
| MUSE-VLParams=7B2025.12 | 72.1 | — | — | — | — | — | 39.7 | 51.3 | 69.8 | 49.6 | — | — | |
| EMU3Params=8B2025.12 | 58.5 | 68.6 | 76.3 | 64.7 | 43.8 | 68.7 | 31.6 | — | 70 | — | 37.2 | — | |
| VILA-UParams=7B2025.12 | — | — | — | 60.8 | — | — | — | — | — | — | 33.5 | — |