List your company

Spatial decision-making on FrozenLake size 16

49.38Success RateSpecFlow
15.923224.609133.29541.9809Jun 1, 2026

Evaluation results

MethodLinks
SpecFlowSetting=Flow(image) + AR(Text)2026.0649.38
Gemini-3-FlashSetting=Naive2026.0646.42
DiffThinkerSetting=Flow(image) Only2026.0643.62
GPT-5.1Setting=Naive2026.0642.85
MVoTSetting=Autoregressive2026.0641.87
Qwen3-VL-32BSetting=SFT2026.0638.24
Qwen3-VL-32BSetting=GRPO2026.0635.91
Qwen3-VL-8BSetting=SFT2026.0629.41
Qwen3-VL-8BSetting=GRPO2026.0626.98
Qwen3-VL-32BSetting=Naive2026.0624.58
Qwen3-VL-8BSetting=Naive2026.0617.21