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

Multimodal Reasoning on RealWorldQA (Acc, Acpt, Speedup)

81.39Accuracy

DREAM-R

8.496427.420746.34565.2693May 27, 2026
Updated 6d ago

Evaluation Results

MethodLinks
2026.05
81.3952.361.8
2026.05
81.3847.431.71
2026.05
78.428.51.72
2026.05
77.8968.052.26
2026.05
76.167.982.48
2026.05
76.125.561.56
2026.05
75.5560.062.19
2026.05
75.36-1.1
2026.05
75.2556.12.03
2026.05
75.2441.66
2026.05
75.17-1.16
2026.05
75.1757.292.13
2026.05
75.1-1.07
2026.05
74.5152.121.61
2026.05
74.2-1.09
2026.05
73.48-1.08
2026.05
73.325.241.4
2026.05
7341.51.52
2026.05
72.9766.852.13
2026.05
72.850.91.93
2026.05
72.5347.221.53
2026.05
72.1249.291.66
2026.05
72.1-1.09
2026.05
71.845.91.52
2026.05
71.1-1.03
2026.05
70.9553.81.82
2026.05
70.539.71.38
2026.05
69.8848.221.48
2026.05
69.8648.511.23
2026.05
69.72-1.12
2026.05
69.0546.91.47
2026.05
68.5244.231.51
2026.05
67.9542.71.49
2026.05
67.8768.751.3
2026.05
62.138.71.26
2026.05
57.135.61.21
2026.05
55.4336.881.23
2026.05
51.623.40.83
2026.05
39.821.351.41
2026.05
11.324.791.09