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

Single-image spatial reasoning on CV-Bench

80.72D Accuracy

Gemini-3-Pro

65.169.1573.277.25Feb 9, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
80.791.285.9
2026.02
77.990.484.2
2026.02
77.184.280.7
2026.02
76.88480.4
2026.02
7586.980.9
2026.02
74.883.379.1
2026.02
7479.576.8
2026.02
73.274.673.9
2026.02
72.274.673.4
2026.02
71.981.176.4
2026.02
71.774.973.3
2026.02
70.879.275
2026.02
70.578.474.5
2026.02
69.372.270.7
2026.02
65.769.567.6