🔍 Read the full analysis: Google Genie 4: The New AI Tool For Creating Walkable Worlds on ThorstenMeyerAI.com
Get the latest gadgets 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 announced Genie 4 on Oct. 27, describing an AI model that generates interactive scenes users can explore and alter with text and controls. The demos are company-selected, the product is not yet generally available, and DeepMind says scenes can drift after a few minutes.
Google DeepMind announced Genie 4 on October 27, presenting it as an AI world model that generates scenes users can move through and change with text while they run. The company says the model can create visuals and sound together, but it is not yet broadly available, and DeepMind acknowledges that objects and details may drift during longer sessions.
In the company’s demonstrations, a user can explore a generated scene with keyboard and mouse controls, then type instructions to alter conditions such as weather or add elements. DeepMind says the system generates the next moments of a scene in response to those inputs rather than relying on a conventional game engine and a stored 3D map. That description is the company’s account of how Genie 4 works.
DeepMind also says Genie 4 produces ambient audio, music and dialogue alongside its images, and can simulate events involving collisions and aerodynamics. It describes Gemini 3.8 Flash as a companion system that interprets goals, breaks them into steps and monitors the resulting frames. The company has not published benchmark figures to quantify these capabilities, and its announcement clips are selected demonstrations rather than independent evaluations.
Access is currently restricted to trusted testers. DeepMind said it plans to offer Genie 4 to Google AI Ultra subscribers through Project Genie and to early partners through the Gemini API, but it gave no launch date or API pricing. DeepMind’s reported tester feedback includes uses such as film previsualization and game prototyping. Its robotics material is a simulation demonstration, not evidence that a physical robot performed the depicted task.
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
A New Tool for Interactive Prototypes
If the announced capabilities hold up outside curated demos, Genie 4 could give creative teams a way to explore rough environments before building them in conventional production tools. A director or game designer might test a scene’s mood, movement or layout by changing prompts while exploring, rather than rendering a fixed clip for every revision. These are potential uses, not established productivity gains: the tool remains in limited testing, and no independent results or cost figures have been published.
The sound component matters because it makes generated scenes more complete as experiences, not merely as images. It may also make the output harder for viewers to distinguish from recorded footage. DeepMind says Genie 4’s video and audio outputs carry an invisible SynthID watermark; the announcement does not specify how reliably that mark can be detected after editing or reposting.
For robotics, simulated environments can support research and testing, but the Genie 4 demo does not show a robot learning or acting in the physical world. DeepMind’s example used a simulated Apptronik Apollo humanoid playing checkers. The distinction between a visual simulation and verified robot performance is important for judging what the release establishes.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Samsung 27-Inch Odyssey OLED G5 QHD Gaming Monitor
Samsung 34-Inch Odyssey G5 Ultra-Wide Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Short Clips to Live Scenes
Genie 4 follows a series of DeepMind world-model releases. The first Genie, introduced in February 2024, generated simple two-dimensional worlds. Genie 2, announced in December 2024, generated three-dimensional environments from a single image and maintained them for up to a minute, according to the company’s description.
In August 2025, DeepMind presented Genie 3 as its first real-time world model, reporting 720p output at 24 frames per second and consistency for a few minutes. Google later made Genie 3 available to paying U.S. subscribers through Project Genie in January 2026. TechCrunch reported a 60-second limit per world, citing compute costs. Those earlier products show the direction of development, but they do not independently verify Genie 4’s claims.
Genie 4 is described as another step toward scenes that respond continuously to user actions. Unlike a finished video, a generated world is meant to keep producing what comes next as a user moves or issues instructions. Unlike a traditional game, DeepMind says, it does not depend on a complete stored map. That distinction helps explain both the appeal of the demonstrations and the risk of visual inconsistency.
“Details and objects “can shift or drift over time.””
— Google DeepMind
Access, Duration and Reliability
Several practical details remain unresolved. DeepMind has not announced when Genie 4 will reach Project Genie subscribers or API partners, how much API access will cost, or what usage limits will apply. The company also has not released benchmark results or an independent assessment of the demonstrations.
DeepMind says sessions can last a few minutes and warns that scene details may drift. TechCrunch and Bloomberg, as cited by German technology publication heise, reported demonstrations lasting up to about two minutes; that figure does not appear in DeepMind’s own announcement text. The available information does not establish a standard session length or how performance varies across prompts and scenes.
It is also unclear how well the system handles complex instructions beyond selected examples, how consistently generated audio matches action, and how the SynthID watermark performs once content is edited. Claims about creative and research uses remain early feedback or demonstrations, not evidence of general commercial readiness.
Testing Before Wider Release
The next stated step is access for trusted testers, followed by planned availability for Google AI Ultra subscribers through Project Genie and early Gemini API partners. Google has not provided dates, so the sequence and pace of any wider rollout are not yet known.
Evidence from broader testing will help show whether Genie 4 can maintain coherent scenes, respond reliably to changing instructions and produce useful synchronized audio beyond handpicked examples. Until Google releases access details and more systematic results, its announcement establishes the intended capabilities and known limitations, but not how well the tool will perform for typical users or production teams.
Key Questions
What is Google Genie 4?
Genie 4 is a world model that DeepMind says generates interactive scenes in response to movement and text instructions. It is designed to produce the next moments of a scene as a user explores, rather than render only a fixed video.
Can the public use Genie 4 now?
No. DeepMind says access is limited to trusted testers for now. It plans access for Google AI Ultra subscribers through Project Genie and early Gemini API partners, but has not given a date or API price.
How long can a Genie 4 world run?
DeepMind describes support for a few minutes of continuous interaction and warns that details can drift. Some media reports put demonstrations at about two minutes, but DeepMind has not stated that figure in its own announcement.
Does Genie 4 control real robots?
The announced robotics example was a simulation involving a virtual Apptronik Apollo humanoid. It does not show a physical robot performing the depicted task.
How does Google identify Genie 4 output?
DeepMind says every Genie 4 video and audio output carries an invisible SynthID watermark. The company has not detailed in the supplied announcement how the mark holds up after editing or redistribution.
Source: ThorstenMeyerAI.com
