This study presents the development and validation of a regional urban building energy model (UBEM) for the island of Gotland, Sweden, using openly available national datasets. The aim is to capture the diversity of the residential building stock-including both urban and rural areas-and provide a robust tool for large-scale energy planning and decarbonisation strategies. The model integrates building geometry, national construction data, and energy performance certificates (EPCs) with probabilistic approaches for infiltration and stochastic occupancy simulation. Building geometry is calibrated against EPC data to assure optimal agreement to real-world circumstances. A novel archetype methodology, based on clustering analysis, is employed to represent the heterogeneous building stock accurately with 15 archetypes for two residential building use types. Implemented with an EnergyPlus-based simulation core, the model achieves high computational efficiency. Validation of aggregated annual results against regional energy use statistics and EPC data demonstrates strong agreement on the aggregate level: for single-family buildings, the annual energy use difference is 3.3%, with a weighted mean difference in energy performance of 0.2 %, while multi-family houses show a modest overestimation. These results confirm that combining open data with advanced probabilistic methods allows reliably simulating building energy dynamics at a regional scale. The framework is easily transferable and adaptable to new case studies, cities, or regions, making it a valuable resource for policymakers and urban planners aiming to enhance energy efficiency and reduce greenhouse gas emissions.