Google's Genie 4 Can Now Build Interactive Prototypes From One Prompt
Type one sentence and get a walkable 3D space with real sound, no studio or coder needed.

Image: Google DeepMind
Quick Summary
- Google DeepMind turned on Genie 4 for Google AI Ultra subscribers on 18 November 2026, inside a new app called Project Genie.
- The model builds a continuous, playable 720p world at 24 frames a second from a text prompt, a photo, or a panorama, complete with real time audio.
- A developer version reached the Gemini API and Google AI Studio on 19 November 2026, letting outside developers build the same world generation into their own apps.
- Every Genie 4 output carries a SynthID watermark and passes safety screening, and a session currently runs for only a few minutes at a time.
Google DeepMind turned on its newest AI model, Genie 4, for real users on 18 November 2026. It builds a continuous, walkable 3D world, complete with real time sound, from nothing more than a sentence or a photo, and developers can now build the same capability into their own apps through the Gemini API.
What did Google DeepMind announce with Genie 4?
Genie 4 is the newest version of Google's world model, an AI system built to generate and run an interactive environment rather than just a flat image or video clip. Google opened it to Google AI Ultra subscribers inside a free standing app called Project Genie on 18 November 2026, with Google AI Pro members able to join a waitlist. A day later, Google for Developers made the same model available through the Gemini API and Google AI Studio, so outside developers can wire world generation into their own products.
Genie 3 and Genie 4 remain world models in the frontier research space. As the research continues to evolve, we are actively tackling key challenges such as camera latency, intricate physical interactions, real time audio precision, and long term scene permanence.from Diego Rivas and Pavan Kumar, Group Product Managers, Google DeepMind and Google Labs (Google, 18 November 2026)
What can Genie 4 actually do?
- Builds a continuous, playable 720p world at 24 frames a second from a text prompt, a reference photo, or a 360 degree panorama image.
- Lets you edit the world mid session: type a new prompt to change the weather, add a character, or trigger an event.
- Generates native audio on the fly, from ambient background sound to character dialogue.
- Runs a session for up to 3 minutes, three times longer than the previous 60 second limit on Genie 3.
- Watermarks every video and audio output with SynthID and runs safety screening on every input and output.
How could your business use it?
None of this was built with a small retailer or clinic in mind, but the tool is open enough to borrow for one. A shop owner could describe a renovated floor layout and walk a landlord or investor through it before a single wall moves. A clinic or restaurant could build a short interactive scenario to train new staff on an awkward customer situation, then edit it live instead of re shooting a training video. An agency could rough out an interactive concept for a client pitch in minutes instead of booking a filming day. Treat it as a fast, rough draft stage, not a finished production tool: a session runs for only a few minutes at a time, and a developer account needs a paid Gemini API key before it earns a place in a real workflow.
Also today
- Canva added a toggle that switches off nine generative AI tools, including Magic Write and Dream Lab, platform wide for free and Pro accounts, giving teams more control over which AI tools touch their brand (PetaPixel, 19 November 2026).
- Tencent published the open weights of its Hy4 Flash model on Hugging Face, a free text and image model with a 1 million token context window that a business can download and run on its own servers (TechNode, 19 November 2026).
- Malaysia's AI Governance Bill is being prepared for enforcement from the day it takes effect, Deputy Digital Minister Wilson Ugak Kumbong said, with the risk based bill due in Parliament early next year (The Star, 17 November 2026).
What should you try this week?
If you already pay for Google AI Ultra, open Project Genie this week and describe one real space from your business, such as your shop floor, your waiting room, or a booth design for your next event. Treat what comes out as a conversation starter for your team or your landlord, not a finished asset. If turning real business ideas into working, deployed tools sounds useful beyond this one demo, that is exactly what NextGen's AI Full-Stack Building for Business (Vibe Coding) program covers.
Common Questions
- What is Google's Genie 4?
- Genie 4 is Google DeepMind's newest world model, an AI system that generates a continuous, playable 3D environment with real time audio from a text prompt or a reference photo. Google opened it to Google AI Ultra subscribers inside a new app called Project Genie on 18 November 2026, then opened a developer version a day later.
- How can a business use Genie 4 right now?
- A Google AI Ultra subscriber can open Project Genie today, describe a space such as a renovated shop floor or a training scenario, then walk through it and edit it live with new prompts. Developers with a paid Gemini API key can build the same world generation into their own apps, per Google's 19 November 2026 developer announcement.
- How much does Genie 4 cost to use?
- Genie 4 is not free. Consumers need a Google AI Ultra subscription to use it inside Project Genie, while Google AI Pro members can only join a waitlist, Google said on 18 November 2026. Developers need a paid tier Gemini API key, and Google points them to its separate pricing documentation rather than listing a flat rate in the announcement.
Source: Google
Sources
- Google DeepMind - Genie 4 is now available (Project Genie) (blog.google)
- Google for Developers - Genie 4 in the Gemini API (blog.google)
- PetaPixel - Canva Now Lets Users Turn Off Generative AI Features (petapixel.com)
- TechNode - Tencent releases Hy4 flash model weights for local deployment (technode.com)
- The Star - AI safety, enforcement framework prepped ahead of new Bill (thestar.com.my)

Written by
Edison Chua
HRD Corp certified AI trainer and founder of NextGen Training Academy, Kota Kinabalu. Trains Malaysian businesses to put AI to work - hands-on, in plain language.
Want To Put This To Work In Your Business?
That's exactly what we teach, hands-on and HRDC-claimable.



