Buildings.city: Scalable urban building energy modeling and carbon emissions mapping using open archetype templates

Cities and municipalities often struggle to scale building-level energy and carbon assessments to the urban scale, due to incomplete data, inconsistent input quality, and a lack of benchmarks. This study presents Buildings.city, an open-source Urban Building Energy Modeling (UBEM) framework and reproducible toolkit for city-scale carbon emissions accounting and mapping. The platform integrates globally available multi-source geospatial data, machine-learning-based archetype inference, and archetype-specific energy simulations (along with the associated templates) to generate detailed building-level carbon maps. To demonstrate the framework's adaptability to diverse urban contexts, the GitHub repository provides a baseline example of Zurich, and the entire toolkit was validated through a city-scale proof of concept in Singapore ( Buildings.sg ). In the full deployment, a predictive model achieved >75% accuracy in inferring missing building archetypes, addressing data gaps in OpenStreetMap (OSM). We also established a set of open-source building energy modeling packages and simulation templates for 23 archetypes, along with a web platform that visualizes simulated carbon emissions across approximately 120,000 buildings to support interactive analysis and policy decision-making. Ultimately, through a modular architecture that integrates diverse datasets, the Buildings.city toolkit provides a transparent, adaptable baseline for universal application. • Developed Buildings.city, a scalable open-source UBEM and carbon mapping toolkit. • Validated via Singapore's first national UBEM templates ( Buildings.sg ). • Machine learning infers missing building archetypes with >75% accuracy. • Jointly models operational and probabilistic embodied carbon for ∼120,000 buildings. • Modular architecture and open templates enable rapid transfer to other cities.