List your company

Visual Reasoning on Manipulation

93.33AccuracyGemini-1.5-Flash
43.4156.3769.3382.29Jun 11, 2026

Evaluation results

MethodLinks
Gemini-1.5-FlashModel Category=Frontier Models2026.0693.33
Gemma-2-9BModel Category=Open-Source Models2026.0690.67
Ours (Stage 2: optimize Lund→gen)Optimization Stage=Stage 2 (Flow-GRPO), Model Category=Our Models2026.0690.67
Seed-1.5-27BModel Category=Frontier Models2026.0689.33
Gemini-1.5-Flash-8BModel Category=Frontier Models2026.0688
Ours (Stage 1: optimize Lgen→und)Optimization Stage=Stage 1 (Reflective SFT), Model Category=Our Models2026.0687.67
Qwen2.5-7BModel Category=Open-Source Models2026.0686.67
ThinkMorphModel Category=Open-Source Models, Interleave Thinking=true2026.0669.33
ThinkMorphModel Category=Open-Source Models, Interleave Thinking=false2026.0665.33
Bagel-7B-MoTModel Category=Open-Source Models2026.0645.33