List your company

Spatial Reasoning on V*

86.9AccuracyQwen3-235B
52.5861.4970.479.31Jun 11, 2026

Evaluation results

MethodLinks
Qwen3-235BModel Type=Open-Source2026.0686.9
Qwen3-32BModel Type=Open-Source2026.0685.8
Qwen3-30BModel Type=Open-Source2026.0684.2
Mini-o3Model Type=Tool-Augmented2026.0682.7
PERIA-8BModel Type=Proposed2026.0682.2
Gemini-2.5-ProModel Type=Commercial2026.0678.5
Qwen3-8BModel Type=Backbone Baseline2026.0677.5
Gemini-2.5-FlashModel Type=Commercial2026.0676.9
w/o ToolModel Type=Ablation2026.0676.4
w/o RLModel Type=Ablation2026.0674.4
InternVL3.5-8BModel Type=Open-Source2026.0671.7
VTool-R1-7BModel Type=Tool-Augmented2026.0670.6
GPT-5Model Type=Commercial2026.0670.1
InternVL3.5-14BModel Type=Open-Source2026.0669.6
w/o SFTModel Type=Ablation2026.0659.2
R1-OneVision-7BModel Type=Tool-Augmented2026.0653.9