Modern high-throughput sequencing at the genomic, transcriptomic, and proteomic levels have improved our understanding of the molecular mechanisms of carcinogenesis. Knowledge about certain molecular alterations further allows individualization of cancer treatment. The pronounced diversity of molecular alterations between tumors - not only between different tumor types, but also between patients with the same tumor type - represents a major challenge in the development of effective treatment strategies. In addition to in-depth molecular characterization of primary tumors, improved cultivation protocols enable the three-dimensional (3D) expansion of tumor cells in vitro, i.e., in the cell culture dish, from a variety of different tumor types. This enables the creation of larger collections, so-called living biobanks, which systematically map the diversity of patient tumors. Through detailed molecular characterization and functional testing with drug screens, these patient-derived 3D tumor models significantly expand the range of tumor models available for preclinical and clinical research. These patient avatars enable a deeper mechanistic understanding of cancer biology and the systematic identification of new vulnerabilities through systematic testing of cancer therapeutics. Therefore, they have enormous potential for the development of stratification approaches in precision medicine, as potential treatment strategies can be tested directly on a patient’s own cells. The use of avatars in oncology thus opens up promising perspectives in personalized cancer treatment.