Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Spatial Reasoning on VSI-Bench

79.2Accuracy

Human

26.57640.23853.967.562Feb 5, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
79.2-------
2026.02
72.6-------
2026.02
68.5-------
67.5-------
2026.02
62.2-------
2026.02
60.9-------
2026.02
60.6-------
2026.02
57.9-------
57.7-------
57.3-------
2026.02
55-------
53.5-------
52.5-------
2026.02
49.9-------
2026.02
49.6-------
49.4-------
2026.02
46.6-------
44.8-------
42.1-------
34-------
32.9-------
31.4-------
29.3-------
28.6-------
2026.02
-27.335.135.933.223.530.130.2
2026.02
-25.931.936.831.522.828.628.7
2026.02
-27.43537.133.723.328.729.8
2026.02
-26.834.43734.42329.531.1
2026.02
-29.436.638.933.528.332.233.4
2026.02
-30.337.739.53431.732.434.6
2026.02
-32.244.542.837.767.847.242
2026.02
-31.943.742.738.668.546.144.1
2026.02
-32.444.542.438.570.746.543.7
2026.02
-3243.84239.467.147.146.8
2026.02
-36.651.248.239.776.149.350.8
2026.02
-37.351.649.940.285.649.252.9