A recent study has compiled an extensive dataset on critical heat flux (CHF) specifically for light-water reactors, encompassing both uniform and non-uniform axial power conditions. This dataset aggregation aims to support Phase II of the OECD/NEA AI/ML CHF benchmark, which focuses on enhancing the accuracy and reliability of thermal-hydraulic operating limits. By addressing non-uniform heating scenarios, the study contributes to a more comprehensive understanding of CHF behavior under realistic reactor conditions. The benchmark itself is designed to leverage artificial intelligence and machine learning techniques to improve safety margins and operational guidelines in nuclear reactor systems. This effort reflects ongoing international collaboration to refine predictive models for CHF, a critical parameter in preventing overheating and potential fuel damage. The study’s dataset serves as a foundational resource for validating and training AI/ML algorithms within the OECD/NEA framework. Overall, this work represents a significant step toward advancing nuclear reactor safety through data-driven methodologies.