MMMU
Massive Multidiscipline Multimodal Understanding benchmark, requiring models to answer questions demanding joint image-text reasoning across 30 disciplines including medicine, law, and the arts. It is one of the core evaluation metrics for assessing a model's general-purpose visual understanding capability.
Timeline (last 90 days)
Qwen3.5 scored 86.7 on TAU2-Bench and 85.0 on MMMU.
MMMU (Massive Multidiscipline Multimodal Understanding) requires models to answer questions that demand joint image-text reasoning across 30 disciplines including medicine, law, and the arts.
Common evaluation benchmarks for visual language models include VQAv2, MMBench, MMMU, TextVQA, and MME.
All Facts (3)
Qwen3.5 scored 86.7 on TAU2-Bench and 85.0 on MMMU.
50%UnverifiedMMMU (Massive Multidiscipline Multimodal Understanding) requires models to answer questions that demand joint image-text reasoning across 30 disciplines including medicine, law, and the arts.
50%UnverifiedCommon evaluation benchmarks for visual language models include VQAv2, MMBench, MMMU, TextVQA, and MME.
50%