Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Multi-step Reasoning on MMMU (Accuracy)

68.3Accuracy (MMMU Multi-step Reasoning)

Claude-3.5

5.3821.71538.0554.385May 13, 2026
Updated 2mo ago

Evaluation Results

MethodLinks
2026.05
68.3
2026.05
67
2026.05
56.7
2026.05
54.8
2026.05
54.3
2026.05
51.9
2026.05
50.8
2026.05
48.8
2026.05
45.2
2026.05
45.1
2026.05
42.9
2026.05
36.6
2026.05
7.8