Skip to content
AI IntelligenceSep 26, 2026AI Intelligence
Article

New benchmark tests AI models on realistic medical records

A new open benchmark evaluates frontier AI models using realistic longitudinal medical records rather than standard exam questions. The testing revealed a significant gap in models' ability to navigate a patient's full history.

Frontier EditorialSource: The Neuron
01

Source Brief

New benchmark tests AI models on realistic medical records: A new open benchmark evaluates frontier AI models using realistic longitudinal medical records rather than standard exam questions. The testing revealed a significant gap in models' ability to navigate a patient's full history.