Google DeepMind launches Genie 4 world model as Project Genie goes live
On November 19, Google DeepMind released the world model Genie 4, calling it its most realistic and interactive world model yet. Project Genie also graduated from preview to a full platform, officially launching with the tagline "Step inside your imagination" for eligible Google AI Ultra subscribers.
Confirmed
- Genie 4's core capability: it fuses real-time physics, reasoning, and native audio, turning text prompts in the browser into real-time interactive worlds within seconds that can be explored and manipulated with a keyboard.
- Real-time editing: from inside the world, users can type new prompts at any time—change the weather, add characters, or alter the scene and storyline—and native audio adapts in real time (a point Sundar Pichai highlighted in his demo).
- New agent: Project Genie introduces an agent that helps users create interactive virtual worlds and build complete experiences around Genie, supporting conversational generation.
- Platform extras: a community gallery lets users discover, remix, and share creations, including remixing pre-generated works.
- Rollout details: Genie 4 is rolling out gradually to eligible Google AI Ultra subscribers; Genie 3 in the model selector was also upgraded, with session length extended to 3 minutes.
- DeepMind also released a behind-the-scenes promo video showing the team navigating scenes with a keyboard and improvising storylines via real-time prompts.
Why it matters
- The Genie series is DeepMind's main line of world-generation models; Genie 4 pushes generated worlds from "watchable" to "real-time intervenable and continuously editable," with native audio plus real-time physics markedly boosting immersion.
- Project Genie leaving preview and building a full creative ecosystem (model + editing + agent + community gallery) signals Google's ambition to turn world models into a mainstream creative platform rather than a pure research demo.
- The subscription gate (Google AI Ultra) means only a small user base initially; how access expands is worth watching.
2026-11-19 ~ 2026-11-20 · 14 related posts
Primary sources
- Google DeepMind Launches Genie 4 World Model and Project Genie for Real-Time Interactive Worlds — Google DeepMind ·
- DeepMind publishes Genie 4 model card: real-time 720p 24fps world generation with audio, on the API — lyraxana ·
- Genie 4 rollout details: Ultra subscribers first, Genie 3 sessions extended to 3 minutes — GoogleDeepMind ·
- [source] Google DeepMind Launches Genie 4 World Model and Project Genie for Real-Time Interactive Worlds — Google DeepMind · 2026-11-19
- Behind the Scenes: Turning Prompts into Real-Time Worlds with Genie 4 — Google DeepMind · 2026-11-19
- Genie 4 arrives in Project Genie with native audio, real-time editing and a creation agent — GoogleDeepMind · 2026-11-19
- [source] Genie 4 rollout details: Ultra subscribers first, Genie 3 sessions extended to 3 minutes — GoogleDeepMind · 2026-11-19
- Google launches full Project Genie ecosystem with Genie 4, real-time editing and native audio — GoogleAI · 2026-11-19
- Google launches Genie 4 world model with real-time editing and native audio — GoogleAI · 2026-11-19
- Google's Project Genie upgrades to Genie 4 with live in-world editing and conversational world building — AI_Andrew · 2026-11-19
- DeepMind launches Genie 4 interactive world model with real-time physics, reasoning and audio — demishassabis · 2026-11-19
- Google Genie 4 rolls out on Project Genie for AI Ultra subscribers — testingcatalog · 2026-11-19
- Google Genie 4 is now available through Project Genie — Recoil42 · 2026-11-19
- Genie 4 Lets You Type New Prompts Mid-Exploration, Audio Adapts in Real Time — sundarpichai · 2026-11-19
- [source] DeepMind publishes Genie 4 model card: real-time 720p 24fps world generation with audio, on the API — lyraxana · 2026-11-20