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

Visual Spatial Reasoning on VSR

88.87Accuracy

Qwen3-VL-32B

52.875662.220371.56580.9097May 9, 2026May 10, 2026May 12, 2026May 13, 2026May 15, 2026May 16, 2026May 18, 2026
Updated 15d ago

Evaluation Results

MethodLinks
2026.05
88.87
2026.05
88.46
2026.05
88.38
2026.05
88.22
2026.05
87.89
2026.05
87.15
2026.05
86.58
2026.05
86.58
2026.05
86.5
2026.05
86.25
2026.05
84.04
2026.05
84.04
2026.05
83.8
2026.05
83.72
2026.05
83.7
2026.05
83.34
2026.05
83.32
2026.05
83.28
2026.05
83.22
2026.05
83.22
2026.05
83.04
2026.05
82.72
2026.05
82.53
2026.05
82.41
2026.05
82.36
2026.05
82.36
2026.05
81.83
2026.05
81.66
2026.05
81.54
2026.05
81.25
2026.05
81.01
2026.05
80.87
2026.05
80.84
2026.05
80.84
2026.05
80.7
2026.05
80.45
2026.05
80.19
2026.05
80.12
2026.05
80.03
2026.05
80.03
2026.05
79.84
2026.05
79.42
2026.05
79.3
2026.05
79.27
2026.05
79.22
2026.05
78.85
2026.05
78.48
2026.05
77.99
2026.05
77.58
2026.05
77.52
2026.05
77.41
2026.05
77.41
2026.05
75.13
2026.05
73.9
2026.05
71.03
2026.05
66.69
2026.05
65.55
2026.05
60.88
2026.05
54.26