description: This lightning talk will present a recently started research project, focusing on energy system model coupling using surrogate models, based on machine learning. Increasing demand for sector-coupled large-energy system models is accompanied by a growing complexity of these models. To limit the complexity, new coupling strategies are required. This project investigates the potential of using machine learning to build surrogate models representing a specific energy sector (e.g. hydrogen) which can then be coupled (e.g. using soft linking) to another operational model (e.g. power system model). The goal is to develop machine learning-based methods to build surrogate models of sectors and link them to existing models.