List your company

Spatial decision-making on Maze size 12

78.31Success RateSpecFlow
29.513242.181654.8567.5184Jun 1, 2026

Evaluation results

MethodLinks
SpecFlowSetting=Flow(image) + AR(Text)2026.0678.31
Gemini-3-FlashSetting=Naive2026.0676.94
DiffThinkerSetting=Flow(image) Only2026.0674.86
MVoTSetting=Autoregressive2026.0672.94
GPT-5.1Setting=Naive2026.0672.83
Qwen3-VL-32BSetting=SFT2026.0658.97
Qwen3-VL-32BSetting=GRPO2026.0655.26
Qwen3-VL-8BSetting=SFT2026.0648.63
Qwen3-VL-8BSetting=GRPO2026.0644.92
Qwen3-VL-32BSetting=Naive2026.0641.92
Qwen3-VL-8BSetting=Naive2026.0631.39