Google DeepMind Unveils Genie 4: A Real-Time World Model With Native Audio

Google DeepMind has introduced Genie 4, the latest iteration of its generative world model, now featuring real-time generation capabilities and native audio support. Genie 4 can generate interactive, navigable environments from prompts, with audio integrated into the world simulation rather than added as a post-processing layer. This represents a significant capability jump for generative world models, which have historically operated without synchronized audio or at speeds too slow for real-time interaction. Developers building game environments, simulation tools, or training data pipelines for robotics will find Genie 4's real-time properties particularly relevant. The native audio dimension also opens possibilities for richer synthetic training environments where sound is a meaningful signal.