ResearchBenchmarksMulti-Agent Reinforcement Learning on SMAC 5m vs 6mFollow90.9Win RateGRASP32.76447.85762.9578.043Jan 22, 2026Feb 2, 2026Feb 14, 2026Feb 25, 2026Mar 9, 2026Mar 20, 2026Apr 1, 2026Evaluation ResultsMethodMethodLinksWin RateStandard DeviationGRASPbase_method=MAPPObase_method=MAPPO2026.0490.9—DG-MAPPODifficulty=Hard, Sight...Difficulty=Hard, Sight Range=92026.0188.74.7MAPPODifficulty=Hard, Sight...Difficulty=Hard, Sight Range=92026.0188.26.2MAT-DecDifficulty=Hard, Sight...Difficulty=Hard, Sight Range=92026.0183.14.6MAPPO2026.0481.5—HAPPODifficulty=Hard, Sight...Difficulty=Hard, Sight Range=92026.0177.57.2MA2E2026.0437.2—HAPPO2026.0435—