List your company

Spatial decision-making on FrozenLake size 4

87.94Success RateSpecFlow
53.900862.737971.57580.4121Jun 1, 2026

Evaluation results

MethodLinks
SpecFlowSetting=Flow(image) + AR(Text)2026.0687.94
Gemini-3-FlashSetting=Naive2026.0686.58
DiffThinkerSetting=Flow(image) Only2026.0684.96
MVoTSetting=Autoregressive2026.0683.31
GPT-5.1Setting=Naive2026.0683.27
Qwen3-VL-32BSetting=SFT2026.0679.58
Qwen3-VL-32BSetting=GRPO2026.0676.43
Qwen3-VL-8BSetting=SFT2026.0671.46
Qwen3-VL-8BSetting=GRPO2026.0668.02
Qwen3-VL-32BSetting=Naive2026.0665.47
Qwen3-VL-8BSetting=Naive2026.0655.21