ResearchBenchmarksNavigation on Physical NAV tasksFollow0.98ASRPPIA0.79280.84140.890.9386Jan 24, 2026Evaluation ResultsMethodMethodLinksASRPPIATarget Model=gemini-1-p2Target Model=gemini-1-p22026.010.98PPIATarget Model=gemini-1-flTarget Model=gemini-1-fl2026.010.98PPIATarget Model=claude-3-...Target Model=claude-3-5-sl2026.010.98PPIATarget Model=gpt-4oTarget Model=gpt-4o2026.010.95PPIATarget Model=gpt-4-turboTarget Model=gpt-4-turbo2026.010.92PPIATarget Model=llama3.2-...Target Model=llama3.2-90b-vi2026.010.8