CityGML has become an international standard for semantic 3D city models for over 15 years, and plays a central role in various applications such as urban planning, environmental analysis, and geospatial infrastructure. The recent release of CityGML 3.0 issued by the Open Geospatial Consortium (OGC) introduces significant enhancements to the data model, which offers higher semantic richness and improved interoperability with IoT and BIM domains for urban digital twins. However, these advancements also necessitate the substantial adaptations of many existing CityGML-compliant software systems. One such system is the 3D City Database (3DCityDB), a widely used open-source geodatabase solution for managing 3D city models. This paper presents the new major version 5.0 of 3DCityDB released in early 2025 and redesigned to provide extensive support for CityGML 3.0 while also preserving compatibility with the earlier CityGML versions. The new 3DCityDB v5 introduces a completely reworked relational schema based on a generic mapping principle, which reduces the structural complexity and improves the extensibility significantly. In addition, a novel approach for geometry storage using database-native spatial types also enhances performance and enables seamless integration with third-party GIS platforms. Moreover, a new command-line interface has been developed to support efficient data importing, exporting, querying, and processing workflows. The paper details the underlying system architecture and implementation strategies, and also presents application scenarios and benchmark results. Future research and development plans are outlined as well.