Genie 2, developed by Google DeepMind, is a large-scale foundational world model capable of generating endless, operable, and playable 3D environments based on a single prompt image, designed for training and evaluating embodied agents. Genie 2 represents a significant advancement in the field of deep learning and artificial intelligence, showcasing various emergent capabilities in large-scale generative models, such as object interaction, complex character animation, and physical simulation, by simulating virtual worlds and their consequences. The research behind Genie 2 fosters new creative workflows for prototyping interactive experiences and opens up new possibilities for the development of future general AI systems and agents.