List your company

Visual Reasoning on OpenWorld

93.6Direct AccuracyQwen2.5-VL 72B
78.83282.66686.590.334Jul 10, 2025

Evaluation results

MethodLinks
Qwen2.5-VL 72BModel size=72B2025.0793.693.286.874.8
Human2025.0791———
Pixtral 12BModel size=12B2025.0779.484.27688