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

Real-world QA on RealworldQA

75.7Accuracy

Qwen2.5-VL-72B

50.01256.68163.3570.019Aug 6, 2024Nov 21, 2024Mar 8, 2025Jun 24, 2025Oct 9, 2025Jan 24, 2026May 12, 2026
Updated 15d ago

Evaluation Results

MethodLinks
75.7
2026.04
75.4
2026.04
74.5
2026.02
73.1
2026.05
72.2
2024.08
71.9
2025.01
71.9
2026.02
71.8
2025.11
71.37
2026.05
71
2025.01
70.2
2025.01
70.2
69.9
2025.01
69.7
2026.02
69.3
2026.05
69
2026.05
68.8
68.5
2026.05
68.2
67.4
2025.12
67.3
2026.05
66.7
2024.08
66.3
2025.11
65.75
2026.05
65.62
2025.12
65.4
2025.11
65.23
2025.01
64.8
2025.12
64.3
2026.05
63.98
2025.01
63.8
2026.05
63.39
2026.05
63.34
2026.05
63.26
2026.05
62.88
2025.12
62.5
2026.05
62.27
2026.05
62.26
2025.12
62.2
2026.05
61.98
2026.05
61.68
2024.08
61.4
2026.05
61.03
2025.01
60.9
2025.11
60.78
2025.01
60.7
2026.04
60.5
60.1
2026.05
59.85
2026.05
59.39
2025.12
58.4
2025.12
58.3
2026.05
58.07
2025.12
57.9
2025.11
57.5
2024.08
55.6
55.6
2026.05
55.42
2025.12
55
2026.05
52.63
2025.12
52.3
2025.12
51