List your company
ResearchFormerly WizWand

Spatial Reasoning on BabyVision

27.3AccuracyGemini-2.5-Pro
6.39611.82317.2522.677Jun 11, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Gemini-2.5-ProModel Type=Commercial2026.0627.3
GPT-5Model Type=Commercial2026.0622.9
Qwen3-32BModel Type=Open-Source2026.0622.9
Qwen3-235BModel Type=Open-Source2026.0621.9
Gemini-2.5-FlashModel Type=Commercial2026.0619.5
PERIA-8BModel Type=Proposed2026.0615.5
Qwen3-30BModel Type=Open-Source2026.0614.9
InternVL3.5-14BModel Type=Open-Source2026.0614.6
R1-OneVision-7BModel Type=Tool-Augmented2026.0614.4
InternVL3.5-8BModel Type=Open-Source2026.0613.9
w/o ToolModel Type=Ablation2026.0613.9
Qwen3-8BModel Type=Backbone Baseline2026.0613
w/o RLModel Type=Ablation2026.0612.4
VTool-R1-7BModel Type=Tool-Augmented2026.0612.3
Mini-o3Model Type=Tool-Augmented2026.0611.6
w/o SFTModel Type=Ablation2026.067.2