🔍 Read the full analysis: How Genie 4 Opens The Door To Walkable AI Worlds on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Google DeepMind presented Genie 4 on October 27, describing it as a real-time model that generates interactive scenes with sound and accepts text changes while they run. The company says access is limited to trusted testers for now; its demos are selected examples, and it acknowledges that worlds can drift and support only a few minutes of continuous interaction.
Google DeepMind demonstrated Genie 4 on October 27, presenting a model that generates interactive scenes users can move through and alter with text while they run. The company says the scenes include native audio and can respond to actions such as keyboard input, but the system is not generally available: access is currently limited to trusted testers, and DeepMind says worlds can drift over a few minutes.
Unlike a conventional video generator, which creates a clip for later viewing, Genie 4 predicts what appears next as a user moves or issues instructions. DeepMind describes this as a world model: it does not use a stored 3D map or a traditional game engine to render the environment. Instead, it generates successive frames based on what has already appeared and on user input. That approach can produce walkable-looking scenes, but can also make details inconsistent as a session continues.
DeepMind says Genie 4 can accept live text instructions to change elements such as weather, camera movement or objects. The company also says it generates sound alongside images, including ambience, music and character dialogue, and simulates physical interactions such as collisions. These capabilities were shown in demonstrations selected by DeepMind; the company has not published benchmark figures in the supplied announcement material.
DeepMind says Gemini 3.8 Flash can observe a scene, break a goal into steps and guide the generated world. A robotics demonstration involved a simulated Apptronik Apollo humanoid playing checkers; the source account says no physical robot was involved. DeepMind also identifies film previsualization and game prototyping as possible uses, based on feedback from testers rather than broad commercial deployments.
Infographic 01 · How Genie grew up
From flat sketches to worlds with sound, in under three years.
Each step as Google DeepMind described it. Session limits and consistency are different measures, so read the labels.
- Feb 2024
Genie 1Simple 2D worlds11 billion parameters, learned from unlabelled internet videos - Dec 2024
Genie 23D worlds from a single imageConsistent “for up to a minute”; real time only as a slimmed-down, lower-quality version - Aug 2025
Genie 3First real-time Genie720p at 24 frames per second, consistent “for a few minutes”, memory about one minute back - Jan 2026
Project GenieUS AI Ultra subscribers can play60 seconds per world, partly because of compute costs - May 2026Project Genie goes globalIn the $200 AI Ultra tier, for users 18 and older
- 27 Oct 2026
Genie 4Sound, live steering, more realismResearch preview for trusted testers; Ultra subscribers and API partners “coming soon”
Genie 3 vs. Genie 4
As described by Google DeepMind. Genie 4 column = vendor claims unless marked.
| Feature | Genie 3 (Aug 2025) | Genie 4 (Oct 2026) |
|---|---|---|
| Sound | Genie 3: None | Genie 4: Generated with the picture: ambient noise, music, dialogue |
| Steering | Genie 3: Keys, plus “promptable world events” | Genie 4: Free-form text changes while it runs; Gemini adapts and checks them |
| Look | Genie 3: 720p, 24 fps | Genie 4: “Unprecedented realism”; no resolution or frame rate published |
| How long | Genie 3: “A few minutes”; Project Genie capped at 60 s | Genie 4: “A few minutes” (DeepMind); up to ~2 min consistent scene reported |
| Access | Genie 3: Small preview, then Project Genie | Genie 4: Trusted testers; Ultra and API partners “coming soon” |
Sources: Genie paper, arXiv, 23 Feb 2024; Genie 2 blog, 4 Dec 2024; Genie 3 blog, 5 Aug 2025; TechCrunch, 29 Jan 2026; Google, 19 May 2026; Google DeepMind, Genie 4 blog, 27 Oct 2026; Genie 4 model page (limitations, disclaimer); TechCrunch, 27 Oct 2026. Table wording: Thorsten Meyer.
Powered by Thorsten Meyer AI
From Generated Clips to Playable Scenes
Genie 4 points to a different way of making interactive environments: people could describe a setting and explore a generated version without developers first building every object and rule in a conventional game engine. If the approach becomes reliable and affordable, it could be useful for early game concepts, film planning and simulated settings for research. Those uses remain prospects, not established outcomes; the announced demos do not show how well the system performs across longer, less controlled sessions.
The audio component also matters. Matching sound and visuals could make generated scenes feel more complete, while making it harder for viewers to identify synthetic footage from appearance alone. The source says Google reports that Genie 4 audio and video outputs carry an invisible SynthID watermark. The material does not explain how reliably that mark can be detected after editing or redistribution, so the statement should not be read as proof that every copy will remain readily identifiable.
Products Worth Considering
Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Logitech MX Mechanical Mini Mac Wireless Keyboard Low-Profile – Space Grey
As an affiliate, we earn on qualifying purchases.
How Genie Has Developed
Genie follows a series of DeepMind systems. The first Genie, introduced in February 2024, generated simple 2D worlds; its research paper described an 11-billion-parameter model trained on unlabeled internet videos. Genie 2, presented in December 2024, generated 3D worlds from a single image and maintained consistency for up to a minute, although its real-time version was a reduced-quality system, according to the supplied account.
Genie 3, announced in August 2025, was described by DeepMind as a real-time model running at 720p and 24 frames per second, with consistency lasting a few minutes and visual memory of about one minute. Google opened Genie 3 through Project Genie to some paying U.S. subscribers in January 2026. TechCrunch reported a 60-second limit per world, attributed partly to computing costs. Those earlier product details provide a comparison point, but do not establish the performance or access terms for Genie 4.
““a few minutes of continuous interaction””
— Google DeepMind
Access, Costs and Session Limits
Genie 4 is available only to trusted testers at present. DeepMind says it is coming to Google AI Ultra subscribers through Project Genie and to early Gemini API partners, but has not supplied a launch date or API price. The source identifies Google AI Ultra as a $200-a-month tier, but does not confirm the subscription terms or geographic availability that will apply when Genie 4 access begins.
It is also unclear how Genie 4 performs outside company-selected demonstrations. DeepMind’s stated limit is a few minutes of continuous interaction; TechCrunch and Bloomberg, as cited by heise, reported scenes of about two minutes, but that figure is not in DeepMind’s own wording. The announcement material provides no independent testing, benchmark results or detailed information on compute requirements. It does not establish whether scenes will remain coherent in longer sessions or how much control users will have over the generated world.
Waiting for Broader Testing
The next concrete development will be access for the groups DeepMind has named: Google AI Ultra subscribers through Project Genie and early Gemini API partners. The company has not announced when that access will start. Wider availability, pricing for API use and any limits on session length remain unconfirmed.
Testing outside the selected demonstrations will help establish whether users can reliably build and revisit scenes, and whether generated sound and physical interactions remain consistent over time. Until those details and independent evaluations are available, Genie 4 is best understood as a demonstrated research and product direction, not evidence that fully stable, open-ended AI worlds are ready for general use.
Key Questions
What is Genie 4?
Genie 4 is Google DeepMind’s world model for generating interactive scenes frame by frame in response to movement and instructions. It differs from a conventional video generator because users can interact with the scene as it is produced.
Can the public try Genie 4 now?
No general public release is described. DeepMind says access is currently limited to trusted testers, with future access planned for Google AI Ultra subscribers through Project Genie and early Gemini API partners. No date has been announced.
Does Genie 4 make fully stable worlds?
DeepMind says objects and details can shift or drift over time and describes continuous interaction as lasting a few minutes. The supplied information does not establish how performance varies across longer sessions or independent tests.
Is a real robot used in the robotics demonstration?
The cited checkers demonstration used a simulated Apptronik Apollo humanoid, according to the source account. It says no physical robot took part.
What might Genie 4 be used for?
DeepMind has cited film previsualization and game prototyping based on tester feedback. These are potential applications, not proof of widespread use or commercial results. The source also describes robot training as a research demonstration.
Source: ThorstenMeyerAI.com
