ResearchBenchmarksMulti-agent Task Fulfillment on AWS Benchmark Mortgage scenarioFollow62.07Goal Success Rate (User)SEAgent49.517252.776156.03559.2939Jan 17, 2026Evaluation ResultsMethodMethodLinksGoal Success Rate (User)Goal Success Rate (System)Goal Success Rate (Overall)User Queries CountExecution Time (s)Token UsageSEAgentMemory=SEMemoryMemory=SEMemory2026.0162.0764.0663.112.7333.993,171.9P2PEnvCommunication=Point-to...Communication=Point-to-point, Memory=Raw history in-context2026.015054.6952.462.4729.685,198.62