ResearchBenchmarksMulti-agent Task Fulfillment on AWS Benchmark Travel scenarioFollow78.79User GSRSEAgent72.487674.123875.7677.3962Jan 17, 2026Evaluation ResultsMethodMethodLinksUser GSRSystem GSROverall GSRUser QueriesExecution Time (s)Token Usage (Count)SEAgentMemory=SEMemoryMemory=SEMemory2026.0178.7969.774.243.1344.383,180.94P2PEnvCommunication=Point-to...Communication=Point-to-point, Memory=Raw history in-context2026.0172.7372.3171.972.6745.854,710.83