Veo & Genie (generative & world models)
Generally Available
Google DeepMind
Overview
DeepMind's generative-media and world-model stack: Veo for high-fidelity video with synchronized audio, and Genie for interactive, navigable 3D worlds. At I/O 2026 these were merged into Gemini Omni, a native world-simulator that understands physical forces like gravity and fluid dynamics, with Project Genie anchored to Google Street View imagery.
Veo 3 (May 2025) added synchronized audio; Gemini Omni Flash launched at I/O 2026 across the Gemini app, Google Flow and YouTube Shorts. Genie world-generation remains partly in research preview.