Dense Multimodal Reasoning on Unifybench++
85.4GPT ScoreGPT-4o+IP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4o+IPModel Size=200B, Method Category=Upper Bound, Input Prompt Type=Image Prompt2026.05 | 85.4 | — | |
| GPT-4o+TPModel Size=200B, Method Category=Upper Bound, Input Prompt Type=Text Prompt2026.05 | 81.9 | — | |
| RAP-MLLMModel Size=13B, Method Category=Und. Only2026.05 | 59.5 | — | |
| Yo’LLaVAModel Size=13B, Method Category=Und. Only2026.05 | 52.9 | — | |
| MC-LLaVAModel Size=13B, Method Category=Und. Only2026.05 | 51.1 | — | |
| Sync-R1Model Size=1.3B, Method Category=Unified Model2026.05 | 50.3 | — | |
| Yo’LLaVA(Phi-1.5)Model Size=1.3B, Method Category=Und. Only2026.05 | 48.4 | — | |
| Yo’ChameleonModel Size=7B, Method Category=Unified Model2026.05 | 39.9 | — | |
| UniCTokensModel Size=1.3B, Method Category=Unified Model2026.05 | 38.5 | — | |
| Chameleon+IPModel Size=7B, Method Category=Unified Model, Input Prompt Type=Image Prompt2026.05 | 37.4 | — | |
| Qwen2.5-VL + TPModel Size=3B, Method Category=Und. Only, Input Prompt Type=Text Prompt2026.05 | 34.1 | — | |
| Show-o+TPModel Size=1.3B, Method Category=Unified Model, Input Prompt Type=Text Prompt2026.05 | 33.7 | — | |
| Chamaleon+TPModel Size=7B, Method Category=Unified Model, Input Prompt Type=Text Prompt2026.05 | 31.3 | — |