🔍 Read the full analysis: Can Google’s Genie 4 Turn A Prompt Into A Walkable World? 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 presented Genie 4 on Oct. 27, describing a model that generates interactive scenes from prompts and responds to controls and text instructions while running. The demonstrations are company-selected, access is limited to trusted testers, and DeepMind says scenes can drift and support only a few minutes of interaction.
Google DeepMind demonstrated Genie 4 on October 27, presenting an AI model that can generate a scene from a text prompt and respond as a user moves through it or requests changes. The company says it produces real-time visuals and audio; public access is not yet available, and DeepMind acknowledges that scenes can drift during interactions.
In the demonstrations, users steer scenes with keyboard and mouse controls and can type instructions while the world is running, such as changing the weather or adding an object. DeepMind describes Genie 4 as a world model: rather than rendering a finished video or drawing from a stored 3D map, it predicts what comes next based on the scene so far and the user’s input.
DeepMind says the model generates sound alongside its imagery, including environmental audio, music and dialogue. It also says a Gemini model can observe the scene, break a goal into steps and help guide what happens next. These capabilities are presented in company-selected demonstrations; DeepMind has not published benchmark figures in the source material.
Access is currently limited to trusted testers. DeepMind said access is coming soon for Google AI Ultra subscribers through Project Genie and for early Gemini API partners. The source gives no release date or API price. TechCrunch and Bloomberg, as cited by heise, report individual scenes lasting up to about two minutes, while DeepMind’s own description says interaction can continue for “a few minutes.”
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 Prompt to Interactive Scene
Genie 4 points toward a different use of generative AI: a prompt could produce a scene that people can explore and alter, rather than a clip they can only watch. If the interaction holds up beyond selected demos, it could support game prototyping and film previsualization by letting creators test settings and actions before building them with conventional tools.
DeepMind reports that testers have used the system for those creative tasks. It has also shown a robotics research experiment in which a simulated Apollo humanoid plays checkers. That is a simulation, not evidence of a real robot performing the task. The distinction matters: a convincing virtual scene does not by itself establish that a model can train robots reliably or replace existing production workflows.
The system also raises questions about synthetic media. DeepMind says Genie 4’s video and audio outputs carry an invisible SynthID watermark. The source does not detail how consistently that watermark can be detected or how it will be surfaced to viewers.
Products Worth Considering
Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Apple MacBook Pro 16-inch Laptop with M5 Pro, 18-Core CPU, 20-Core GPU, 48GB Memory, 2TB SSD, Space Black
As an affiliate, we earn on qualifying purchases.
How Genie Reached Real Time
Google’s Genie line has progressed from simpler generated environments toward interactive scenes. The first Genie, introduced in 2024, generated simple 2D worlds. Genie 2 followed with 3D worlds based on a single image, but its more capable experience was not fully real-time. 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.
Project Genie later offered Genie 3 to paying U.S. subscribers, according to the supplied source, with TechCrunch reporting a 60-second limit per world. That history helps explain why duration, consistency and computing cost remain central questions for Genie 4. DeepMind’s description of the new model still acknowledges that objects and details can shift over time.
Genie 4 is not described as having a conventional game engine or a persistent 3D map underneath. Its scenes are generated from learned patterns and prior frames. That approach can create visually rich environments from a prompt, but also means the model may not preserve every detail when a user turns around or continues interacting.
“Details and objects “can shift or drift over time.””
— Google DeepMind
Limits Behind the Demonstrations
Genie 4 has not been made generally available, so people outside the tester group cannot independently assess its performance. The source says DeepMind selected the clips it showed, and the company has not supplied benchmark results. The demonstrations therefore do not establish how reliably the system handles varied prompts, long sessions or unexpected user actions.
There is also a difference between DeepMind’s description of interaction lasting “a few minutes” and media reports of scenes reaching about two minutes. The exact session limit and how it varies by access tier are not confirmed in the company’s stated material. Details of pricing, API availability and launch timing remain unannounced. It is also unclear how often world details drift in ordinary use or whether SynthID watermarks will be detectable in all viewing conditions.
Access and Testing Ahead
The next milestone is access for Google AI Ultra subscribers through Project Genie and early Gemini API partners, which DeepMind has described as coming soon. The company has not announced a date, API price or broader rollout plan. Any wider release would allow more users to test the model’s scene consistency, interaction length and audio generation outside curated demonstrations.
Until those details arrive, Genie 4 remains a demonstrated system rather than a generally available creative tool. Further information from Google on access terms, performance testing and the limits of continuous interaction will clarify how close it is to practical use.
Key Questions
What is Google Genie 4?
Genie 4 is a Google DeepMind world model that generates a scene and predicts what happens next as a user moves or enters instructions. It is not described as a conventional game engine with a stored 3D map.
Can the public use Genie 4 now?
No. DeepMind says access is limited to trusted testers, with availability for Google AI Ultra subscribers through Project Genie and early Gemini API partners described as coming soon. No date has been provided.
How long can users interact with a Genie 4 world?
DeepMind says the model supports a few minutes of continuous interaction. TechCrunch and Bloomberg, as cited by heise, report scenes of up to about two minutes; the exact limit has not been confirmed by DeepMind in the supplied material.
Does the robotics demonstration show a real robot using Genie 4?
No. The cited demonstration involved a simulated Apollo humanoid playing checkers. It does not show a physical robot completing the task.
What are Genie 4’s main limitations?
DeepMind says scene details and objects can shift or drift over time, and interaction lasts only a few minutes. The company has not published benchmark results in the source material, and public users have not yet been able to test the system.
Source: ThorstenMeyerAI.com
