ResearchBenchmarksNavigation on MiniGrid 3 uncertain objectsNavigation on MiniGrid 3 uncertain objectsFollow1,000Success RateMINT59.84303.92548792.08Feb 4, 2026Evaluation resultsMethodMethodLinksSuccess RateAverage RewardMINTVersion=LimitedVersion=Limited2026.021,000—o3-miniMethod Category=LLMMethod Category=LLM2026.02100—o3Method Category=LLMMethod Category=LLM2026.02100—GPT-4oMethod Category=LLMMethod Category=LLM2026.0296—