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

Visual Search on HR-Bench 4K

79.4Accuracy

Qwen2.5-VL-72B

52.98459.84266.773.558Dec 19, 2025Jan 14, 2026Feb 9, 2026Mar 8, 2026Apr 3, 2026Apr 29, 2026May 26, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2025.12
79.4
78.5
2026.02
78.1
2026.02
77.9
77.9
2026.02
77.5
2025.12
77
2026.02
77
2026.05
77
2026.02
76.1
2026.02
75.6
2025.12
75.5
2025.12
75.2
2026.02
75.2
2026.02
75.1
2025.12
74.8
2025.12
74.8
2026.02
74
2025.12
72.9
2025.12
72
2026.02
71.6
2025.12
70.5
70
2025.12
69
68.8
2025.12
68.5
2025.12
66.3
2025.12
65
54