Focusing on ultra-high voltage backbone grids, the application of large model technologies in this field is reviewed. First, the challenges faced by high-voltage direct current (HVDC) systems in achieving secure, stable, and efficient operation and maintenance under the energy transition are elaborated, and the solutions provided by large models are clarified. Second, the key characteristics of HVDC systems and the fundamental principles of large models in smart grids are summarized, with a focus on key technologies tailored to HVDC applications, including the construction of equipment knowledge systems, lightweight modelling of converter stations, and data-secure collaborative architectures. Then, through practical case studies, the applications and effectiveness of large models in HVDC fault diagnosis and prediction, operational evaluation and dispatch, and asset management are illustrated. Finally, the issues associated with large model applications, such as data quality and security, interpretability, and computational costs are analyzed, and future research directions and prospects are discussed. The study shows that large models can facilitate the transition of HVDC operation and maintenance from reactive repair to predictive maintenance, and enable the evolution of dispatching from experience-based decision-making to intelligent decision-making, which is of great significance for ensuring the safe and efficient operation of HVDC systems.