🔍 Read the full analysis: What Genie 4 Gets Right—and What Still Breaks on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Google DeepMind announced Genie 4 on October 27, describing a real-time world model that responds to player controls and text while generating visuals and audio. The company says scenes can still drift after a few minutes; access is limited to trusted testers, with no public release date or independent benchmark results provided.
Google DeepMind announced Genie 4 on October 27, presenting it as a real-time world model that can generate interactive scenes, respond to typed instructions and produce sound alongside images. The announcement matters because it advances Google’s effort to make AI-generated environments usable for games, film planning and research, but the company says scenes can still shift or drift over time and support only a few minutes of continuous interaction.
Unlike a video generator that renders a finished clip from a prompt, Genie 4 is designed to predict each new frame as a user moves through a scene or issues a change. DeepMind says users can steer environments with keyboard and mouse controls, then type instructions to alter elements such as weather, objects or camera movement. The system does not rely on a stored 3D map, according to the source account; it generates what comes next from the scene and inputs so far. That approach can produce responsive environments, but also leaves room for details to change as a session continues.
DeepMind describes more realistic visuals and physics, people that behave more naturally, and audio generated alongside the scene, including environmental sounds, music and dialogue. It also says Gemini 3.8 Flash can act as a guiding agent, breaking goals into steps and monitoring generated frames. These are company claims and selected demonstrations: DeepMind did not publish benchmark figures in the supplied material, and there is no independent performance evaluation cited there.
Access remains restricted to trusted testers. DeepMind says Genie 4 is coming to Google AI Ultra subscribers through Project Genie and to early Gemini API partners, but has not announced a launch date or API price. The source identifies Google AI Ultra as a $200-a-month tier. Reported uses include film previsualization and game prototyping based on tester feedback; a robotics example is described as a research simulation, not a test involving a physical robot.
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 Interactive Worlds
Genie 4’s significance lies in the shift from generating a scene once to responding continuously to user input. If the system becomes reliable and accessible, creators could use it to explore environments or test ideas without building every asset and interaction in a conventional game engine. Researchers could also use simulated environments to examine possible scenarios, although the source material does not establish that Genie 4 is ready for operational training.
The limits are relevant to those potential uses. A world that changes its details over time may be difficult to use for precise planning, repeatable tests or long-form production. Its current short interaction window also constrains how much work can be done in one session. DeepMind’s demonstrations indicate a direction for the technology, but they do not establish that it can replace established production tools or provide dependable simulations for safety-critical work.
Audio adds another consideration for audiences and creators: synthetic scenes may appear and sound more complete. Google says Genie 4 video and audio outputs include an invisible SynthID watermark. The announcement does not explain how consistently that watermark can be detected after content is edited or shared, so it should not be treated as a complete answer to questions about identifying generated media.
Products Worth Considering
Logitech MX Mechanical Full-Size Wireless Mechanical Keyboard, Tactile Quiet Switches, Backlit Keys, Bluetooth, USB-C, Multi-Device, Graphit
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Genie Reached Real-Time Worlds
Genie’s earlier versions show the progression behind the announcement. The first model, introduced in February 2024, generated simple 2D worlds from video training data. Genie 2, announced in December 2024, created 3D environments from a single image and maintained them for up to a minute, though real-time use required a reduced-quality version. In August 2025, DeepMind presented Genie 3 as its first real-time world model, running at 720p and 24 frames per second and maintaining consistency for a few minutes, with visual memory reaching back about one minute, according to the source.
Google opened Genie 3 through Project Genie to paying U.S. subscribers in January 2026. TechCrunch reported a 60-second limit per world at that time, attributing it partly to computing costs. That figure concerns the earlier service and should not be confused with Genie 4’s stated limit of a few minutes. TechCrunch and Bloomberg, as cited by heise, reportedly described Genie 4 clips of about two minutes, but that duration is not in DeepMind’s own wording.
“Details and objects “can shift or drift over time.””
— Google DeepMind
Access, Benchmarks and Session Length
Several details remain unsettled. DeepMind has not given a public release date, API pricing or benchmark results in the material provided. It is also unclear how Genie 4 performs across longer sessions, varied prompts or repeated runs beyond the selected company demonstrations. DeepMind’s phrase “a few minutes” is not an exact limit; the roughly two-minute figure reported by other outlets is not stated in the company’s own announcement.
The practical status of the use cases also varies. Film previsualization and game prototypes are described as feedback from testers, while the robotics example is a simulation involving a virtual Apollo humanoid and a checker-playing task. It does not show a physical robot performing the task. The source says Waymo uses a simulator based on the predecessor Genie 3, but does not establish that Genie 4 is being used for Waymo operations.
The Next Test Is Wider Access
The next stated step is access for Google AI Ultra subscribers through Project Genie and early partners using the Gemini API. DeepMind has not specified when either group will be able to use Genie 4 or whether access will expand beyond them. Until availability and independent testing are clearer, the most useful evidence will be whether users can keep scenes coherent across longer sessions and whether the system’s controls work reliably beyond curated demos.
Google’s watermarking claim will also draw attention as synthetic video and audio become more integrated. The announcement says outputs carry SynthID, but does not detail detection performance or how the marker holds up through common editing workflows. Those questions, along with reliability, cost and session limits, will shape whether Genie 4 moves from a demonstration and research tool into routine creative or technical use.
Key Questions
What is Genie 4?
Genie 4 is a real-time world model from Google DeepMind. It generates the next part of a scene as users move through it or type instructions, rather than producing only a finished video clip.
Can the public use Genie 4 now?
No general public release is confirmed. DeepMind says access is coming to Google AI Ultra subscribers through Project Genie and early Gemini API partners, but has not announced a date or API price.
How long can a Genie 4 scene run?
DeepMind describes support for a few minutes of continuous interaction and says objects can drift over time. Reports of scenes lasting about two minutes come from other outlets, not the company’s own wording.
Does Genie 4 control a real robot?
The robotics example in the source is a simulation of an Apptronik Apollo humanoid playing checkers. It does not show a physical robot performing the task.
How does Google say people can identify Genie 4 media?
Google says Genie 4 video and audio outputs carry an invisible SynthID watermark. The announcement does not specify how reliably it can be detected after editing or redistribution.
Source: ThorstenMeyerAI.com
