List your company

Visual Reasoning on VisualProbe (test)

47.2Accuracy (Hard)Mini-o3
8.92818.86428.838.736Dec 19, 2025Jan 12, 2026Feb 6, 2026Mar 2, 2026Mar 27, 2026Apr 20, 2026May 15, 2026

Evaluation results

MethodLinks
Mini-o3Source=Reproduced through available model weights2025.1247.245.264.550.9
DRIM2025.1245.348.169.553.4
DeepEyesSource=Reproduced through available model weights2025.1237.831.366.742.3
GPT-52025.1231.12257.533.6
Gemini 2.5 Pro2025.1229.336.653.239.6
GLM-9B-DeltaThinkerAlgorithm=DeltaPrompts2026.0520.0421.5550.04—
REVisual-R12026.0518.8720.1543.97—
Qwen-8B-DeltaThinkerAlgorithm=DeltaPrompts2026.0516.9820.1551.77—
GLM-4.1V-9B-Thinking2026.0516.0417.9139.01—
Vision-R1-7B2026.0515.4716.436.03—
LLaVA-OneVisionSource=Reported in original or other relevant paper2025.1213.412.536.2—
Ovis2.5-9B2025.1212.326.951.131
ARES-RL-7B2026.0511.2418.538.94—
Qwen3-VL-8B-Thinking2026.0510.8613.6134.75—
Bee-8B-RL2026.0510.414.9326.95—