🔍 Read the full analysis: Google’s Genie 4 Brings AI-Generated Worlds Within Reach on ThorstenMeyerAI.com
Get hardware and tech essentials 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 demonstrated Genie 4 on October 27, describing it as a real-time world model that generates interactive scenes, sound and changes based on text instructions. The company has not released it broadly, published benchmark figures or set an access date; DeepMind says scenes can drift and support only a few minutes of interaction.
Google DeepMind demonstrated Genie 4 on October 27, presenting it as a real-time AI model that creates interactive scenes with generated sound and can respond to keyboard, mouse and text input. The system is not yet broadly available: DeepMind says it is being tested by selected users and partners, and acknowledges that details can drift during sessions that last only a few minutes.
In DeepMind’s demonstration, a user can move through a generated scene using standard controls and type requests to change it, such as altering the weather. The company says Genie 4 generates the next moment from the scene so far and from user input, rather than rendering a finished video or drawing from a stored 3D map. It describes the system as a world model, designed to predict and generate what happens next as a scene is explored.
DeepMind says the model creates audio alongside images, including environmental sounds, music and dialogue, and can simulate events such as collisions and aerodynamics. It also says a Gemini model can monitor scenes and help break tasks into steps. These are company descriptions supported by its selected demonstrations; DeepMind has not published benchmark figures in the supplied announcement material. The examples do not independently establish how the system performs across other prompts or settings.
Access remains restricted. DeepMind says the model is available to trusted testers and is expected to come to Google AI Ultra subscribers through Project Genie and early Gemini API partners. It has not provided a launch date or an API price. DeepMind also says every Genie 4 video and audio output carries an invisible SynthID watermark.
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
Where Interactive AI Could Be Used
If the demonstrations translate into dependable tools, Genie 4 could let creative teams test settings and interactions without building every element of a conventional game or film environment first. DeepMind says testers have reported using world models for film previsualization and game prototypes. That is early feedback, not evidence of established production use or measurable savings.
The potential also reaches beyond entertainment. DeepMind showed a robotics research scenario in which a model broke a task into steps and Genie 4 simulated a humanoid robot playing checkers. The source material says no physical robot was involved in that demonstration. It should be read as a simulation example, not proof that Genie 4 can train robots reliably in real-world conditions.
Generated scenes that combine realistic-looking images, motion and sound may also make it harder for viewers to judge whether a clip depicts a real event. DeepMind’s stated use of SynthID addresses identification at the output level, but the announcement does not explain how viewers will check for the watermark or how well it survives reposting and editing. The technology’s practical impact will depend on both its reliability and the safeguards attached to public access.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit
LG UltraGear 45GX900A-B 45-Inch Curved OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Genie 3 to Live Scenes
Genie 4 follows earlier versions that increased the complexity and interactivity of generated environments. The first Genie, introduced in 2024, created simple 2D worlds. Genie 2 later generated 3D environments from a single image and maintained them for up to a minute, according to the supplied account. In 2025, DeepMind described Genie 3 as its first world model to run in real time, at 720p and 24 frames per second, with consistency lasting a few minutes.
Google made Genie 3 available to some paying US subscribers through Project Genie in January 2026, according to the source material. TechCrunch reported a 60-second limit per world for that release and linked it partly to computing costs. That figure refers to the earlier product and is not a published Genie 4 limit. For Genie 4, DeepMind’s own wording is that it supports a few minutes of continuous interaction; TechCrunch and Bloomberg, as cited by heise, reported demonstrations of scenes lasting about two minutes.
Unlike a conventional game engine, a world model generates a scene as it goes rather than retrieving every object and surface from a prebuilt map. That approach can produce varied environments, but it also creates a consistency challenge: the model may not reproduce a detail the same way after the user moves away and returns. DeepMind itself warns that objects and details can shift or drift over time.
“Details and objects can shift or drift over time.”
— Google DeepMind
Limits Behind the Demonstrations
Genie 4’s access date, Gemini API pricing and the terms for testers have not been announced. DeepMind has also not supplied public benchmark results in the source material, so comparisons with other models on realism, response speed, sound quality or session stability cannot be assessed from the demonstrations alone.
The examples were selected by DeepMind, and outside users cannot yet test whether the results hold across a broad range of instructions. The company acknowledges that objects can drift and gives a general interaction duration of a few minutes. Reports of scenes lasting about two minutes come from TechCrunch and Bloomberg as cited by heise, rather than from a precise runtime stated in DeepMind’s own text.
Several use cases remain at an early stage. Film and game applications are described as tester feedback reported by DeepMind, while the robotics example was a simulation rather than a test with a real robot. It is also not yet clear how effectively SynthID can be used to identify generated material after distribution or editing.
Access and Independent Testing
The next public milestone is likely to be wider access through Project Genie and early Gemini API partnerships, but DeepMind has not set a date. Any release details will clarify who can use the model, how long sessions can run, what it costs through the API and whether use is limited by geography or subscription tier.
Independent testing will be needed to establish how reliably Genie 4 follows instructions, maintains objects and locations, synchronizes sound with events and handles longer interactions. Until those tests and access terms are available, its capabilities are best understood as company-demonstrated features, with performance beyond the selected clips still unsettled.
Key Questions
What is Google Genie 4?
Genie 4 is a world model that generates an interactive scene as a user moves through it or gives instructions. DeepMind says it creates images and audio in real time rather than producing only a finished video.
Can the public use Genie 4 now?
No broad public release has been announced. DeepMind says access is limited to trusted testers, with Google AI Ultra subscribers through Project Genie and early Gemini API partners expected to get access later. There is no announced date.
How long can a Genie 4 scene last?
DeepMind describes Genie 4 as supporting a few minutes of continuous interaction. Reports of scenes around two minutes come from TechCrunch and Bloomberg as cited by heise; DeepMind has not specified an exact maximum in the supplied material.
What are Genie 4’s main limitations?
DeepMind says details and objects can shift or drift over time, and that interaction lasts only a few minutes. The company has not published benchmark figures in the source material, and the public has not had broad access for independent testing.
Does Genie 4 generate sound, and how are outputs marked?
DeepMind says Genie 4 generates environmental audio, music and dialogue with its scenes. The company says every video and audio output includes an invisible SynthID watermark; details about how viewers can verify it are not provided in the supplied announcement.
Source: ThorstenMeyerAI.com
