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

Multimodal Understanding on MMMU, SEED, OCRBench, VizWiz, ScienceQA, and TextVQA (test)

78.1Average Accuracy

FP16

23.39637.59851.866.002May 20, 2026
Updated 8d ago

Evaluation Results

MethodLinks
2026.05
78.1
2026.05
77.8
2026.05
77.6
2026.05
77.5
2026.05
77.5
2026.05
76.6
2026.05
75.8
2026.05
75
2026.05
74.6
2026.05
74.3
2026.05
74.2
2026.05
74
2026.05
73.9
2026.05
73.9
2026.05
73.8
2026.05
73.6
2026.05
73.6
2026.05
73.5
2026.05
73.3
2026.05
73.1
2026.05
72.6
2026.05
72.4
2026.05
72.3
2026.05
72.3
2026.05
72.1
2026.05
71.9
2026.05
71.7
2026.05
71.4
2026.05
71.3
2026.05
71.1
2026.05
71.1
2026.05
70.9
2026.05
70.9
2026.05
70.6
2026.05
70.4
2026.05
70.3
2026.05
70.2
2026.05
70
2026.05
69.8
2026.05
69
2026.05
68.2
2026.05
67.9
2026.05
67.3
2026.05
67.2
2026.05
67.1
2026.05
67
2026.05
66.9
2026.05
66.7
2026.05
66.5
2026.05
66.5
2026.05
66.3
2026.05
66.3
2026.05
65.8
2026.05
65.4
2026.05
65.3
2026.05
65.1
2026.05
64.8
2026.05
64
2026.05
63.4
2026.05
62.4
2026.05
61.5
2026.05
61.4
2026.05
61.3
2026.05
60.4
2026.05
60.2
2026.05
59.6
2026.05
59.6
2026.05
59.4
2026.05
59.1
2026.05
58.9
2026.05
55.1
2026.05
53
2026.05
52.5
2026.05
51.8
2026.05
50.5
2026.05
48.5
2026.05
47.9
2026.05
46.6
2026.05
44.3
2026.05
42.2
2026.05
42.2
2026.05
40.4
2026.05
38
2026.05
25.5