ResearchBenchmarksMulti-agent system security evaluation on LANGMANUS (test)Follow0ASRCrescendo-0.001-0.000500.0005Dec 3, 2025Evaluation ResultsMethodMethodLinksASRCrescendoModel=All Models (GPT-...Model=All Models (GPT-4O-1120, QWEN-VL-MAX, DOUBAO-VISION-PRO), System Type=Multi-Agent2025.120Pop-upModel=All Models (GPT-...Model=All Models (GPT-4O-1120, QWEN-VL-MAX, DOUBAO-VISION-PRO), System Type=Multi-Agent2025.120