Recent advancements in radiation therapy have significantly improved treatment precision and patient outcomes. However, the complexity of biological responses to radiation—ranging from subcellular damage to systemic effects—remains a major challenge in designing truly personalized treatment regimens. The emerging concept of multiscale digital twins offers a transformative approach by integrating computational modeling with patient-specific data to simulate tumor and normal tissue responses to radiation. These predictive simulations enable real-time treatment adaptation and optimization, paving the way for more effective, individualized radiation therapy. However, the full potential of digital twins in radiation oncology depends on high-performance computational platforms capable of delivering these models at clinically relevant timescales. This presentation will explore recent developments in multiscale digital twin frameworks for radiation therapy, highlighting key challenges, computational requirements, and potential clinical applications in optimizing treatment protocols.