List your company
ResearchFormerly WizWand

Visual Probing on VisualProbe Easy

70.2AccuracyMini-o3
17.05630.85344.6558.447Jun 11, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Mini-o3Model Type=Tool-Augmented2026.0670.2
GPT-5Model Type=Commercial2026.0662.4
Gemini-2.5-ProModel Type=Commercial2026.0662.1
PERIA-8BModel Type=Proposed2026.0661.7
Qwen3-235BModel Type=Open-Source2026.0658.8
InternVL3.5-14BModel Type=Open-Source2026.0658.1
Qwen3-32BModel Type=Open-Source2026.0657.4
InternVL3.5-8BModel Type=Open-Source2026.0654.7
Qwen3-8BModel Type=Backbone Baseline2026.0654.6
Qwen3-30BModel Type=Open-Source2026.0651.7
w/o RLModel Type=Ablation2026.0651
Gemini-2.5-FlashModel Type=Commercial2026.0646.1
R1-OneVision-7BModel Type=Tool-Augmented2026.0639.2
w/o ToolModel Type=Ablation2026.0636.1
VTool-R1-7BModel Type=Tool-Augmented2026.0630.5
w/o SFTModel Type=Ablation2026.0619.1