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

Video Understanding on MLVU (dev)

68.4MLVU Dev Score

Full Model (Qwen2.5-Omni-7B)

31.79241.29650.860.304Mar 20, 2025May 29, 2025Aug 8, 2025Oct 18, 2025Dec 28, 2025Mar 9, 2026May 19, 2026
Updated 14d ago

Evaluation Results

MethodLinks
2025.12
68.4--
2025.12
68.3--
2025.12
67.4--
2025.03
67.2--
2025.12
67.2--
2025.03
66.7--
2025.12
66.1--
2025.03
65.3--
2025.03
65.1--
2025.12
65--
2025.12
63.4--
2025.12
62.9--
2025.03
62.8--
2026.05
59.71--
2026.05
59.61--
2025.03
56.3--
2026.05
53.84--
2025.03
48.5--
2026.05
47.91--
2025.03
47.3--
2025.03
33.2--
2024.12
-64.6-
2024.12
-60.5-
2024.12
-60.9-
2024.12
-66.4-
2024.12
-30.2-
2024.12
-58.6-
2024.12
-60.2-
2024.12
-64.7-
2024.12
-64.7-
2026.02
--35.5
2026.02
--36.4
2026.02
--41.9
2026.02
--42.2
2026.02
--44.5
2026.02
--44.5
2026.02
--46.4
2026.02
--48.5
2026.02
--47.3
2026.02
--48.9