ResearchBenchmarksData Analysis Question Answering on AggQA MedicalFollow90.91Accuracy (Easy)PISA15.270834.907954.54574.1821Oct 12, 2025Evaluation ResultsMethodMethodLinksAccuracy (Easy)Accuracy (Medium)Accuracy (Hard)Overall AccuracyPISA2025.1090.919653.8583.67LangMem2025.1072.73647.6951.02FullContextBackbone=GPT-5 MiniBackbone=GPT-5 Mini2025.1045.466823.0851.02AMem2025.1027.2728020.41Mem02025.1027.274015.3830.61MemoryOS2025.1018.1812010.2