Google DeepMind releases Genie 4 world model with native audio and roughly two-minute experiences
Google DeepMind released the Genie 4 world model on October 27 as a research preview for trusted testers, before coming to Google AI Ultra subscribers through Project Genie and early access partners of the Gemini API and Google AI Studio. It runs on Gemini 3.8 Flash and produces experiences about two minutes long, up from roughly one minute on Genie 3, and for the first time natively generates ambient sound and interactive dialogue. Users can reshape weather, lighting, camera and scene elements in real time with natural language while Gemini keeps watching frames and verifying progress step by step. Google acknowledges that visual detail, object permanence, audio accuracy and spatial consistency all degrade as simulations run longer or after repeated edits, and all generated video and audio carries an invisible SynthID watermark.