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

Spatial Reasoning (Video) on VSI-Bench

79.2Accuracy

Human

24.0838.3952.767.01Oct 10, 2025Oct 30, 2025Nov 20, 2025Dec 11, 2025Dec 31, 2025Jan 21, 2026Feb 11, 2026
Updated 7d ago

Evaluation Results

MethodLinks
2025.10
79.2------
2026.02
68.355.280.565.153.687.445.2
2026.02
67.153.279.364.451.877.245
2025.10
66.3------
2026.02
62.548.775.962.249.466.343.3
2026.02
58.644.576.559.14870.242.7
2026.02
57.746.177.459.747.968.444.9
2026.02
56.447.776.358.747.267.444.1
2026.02
52------
2025.10
48.6------
2026.02
48.4------
2025.10
48.4------
2025.10
46.9------
2025.10
46.3------
2025.10
46.1------
2026.02
45.6------
2026.02
45.4------
2025.10
45.4------
45------
2025.10
44.2------
2026.02
42.1------
2025.10
42------
2025.10
38.2------
2025.10
35.6------
2026.02
34.7------
2025.10
32.7------
2026.02
32.4------
2025.10
32.4------
2025.10
30.7------
2026.02
26.2------