What Google’s Genie 4 Can Do, Where It Struggles, And Who Can Try It
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: What Google’s Genie 4 Can Do, Where It Struggles, And Who Can Try It on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the little things that make your day delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Google DeepMind introduced Genie 4 on October 27, describing it as a real-time world model that generates interactive scenes, sound and changes prompted by text. The company says access is coming to selected testers, but has not given a date; it also acknowledges that scenes can drift and support only a few minutes of interaction.

Google DeepMind announced Genie 4 on October 27, describing it as a real-time AI world model that generates interactive scenes with sound and responds to keyboard, mouse and text prompts. The system is not generally available: DeepMind says it is coming first to trusted testers, including Google AI Ultra subscribers through Project Genie and selected Gemini API partners, without giving a release date.

Unlike a conventional video generator that renders a clip from a prompt, Genie 4 is designed to produce the next moment of a scene as a person moves through it or gives new instructions. DeepMind says users can steer scenes with standard controls and prompts to change conditions such as weather, add people or objects, or alter the camera. The company says its Gemini model helps interpret instructions and monitor what happens in the generated scene.

DeepMind also describes natively generated audio, including ambient sounds, music and character dialogue, alongside more realistic people and simulated physics. Those capabilities are presented in company-selected demonstrations, not independently verified benchmarks. DeepMind has published no benchmark figures in the source material. It says the model can simulate events such as collisions and aerodynamics, but specific performance measures and comparisons are not provided.

The company acknowledges two practical limits: objects and scene details can shift or drift over time, and the model supports “a few minutes” of continuous interaction. TechCrunch and Bloomberg, as cited by heise, report demonstrations of up to about two minutes, but that duration is not stated in DeepMind’s own description. DeepMind says video and audio outputs carry an invisible SynthID watermark.

At a glance
announcementWhen: Announced October 27, 2026; access desc…
The developmentGoogle DeepMind announced Genie 4, a world model for creating and steering interactive, sound-enabled scenes, and outlined limited upcoming access.
Thorsten Meyer AI
Research preview · 27 Oct 2026

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.

4Genie generations, February 2024 to October 2026
60 sTime limit per world in Project Genie with Genie 3 (TechCrunch, Jan 2026)
≈2 minGenie 4 consistent scene, as reported by TechCrunch; not in DeepMind’s own text
$200Monthly price of the AI Ultra tier that includes Project Genie (US, since May 2026)
  • Feb 2024
    Genie 1
    Simple 2D worlds11 billion parameters, learned from unlabelled internet videos
  • Dec 2024
    Genie 2
    3D worlds from a single imageConsistent “for up to a minute”; real time only as a slimmed-down, lower-quality version
  • Aug 2025
    Genie 3
    First real-time Genie720p at 24 frames per second, consistent “for a few minutes”, memory about one minute back
  • Jan 2026
    Project Genie
    US 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 4
    Sound, 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.

FeatureGenie 3 (Aug 2025)Genie 4 (Oct 2026)
SoundGenie 3: NoneGenie 4: Generated with the picture: ambient noise, music, dialogue
SteeringGenie 3: Keys, plus “promptable world events”Genie 4: Free-form text changes while it runs; Gemini adapts and checks them
LookGenie 3: 720p, 24 fpsGenie 4: “Unprecedented realism”; no resolution or frame rate published
How longGenie 3: “A few minutes”; Project Genie capped at 60 sGenie 4: “A few minutes” (DeepMind); up to ~2 min consistent scene reported
AccessGenie 3: Small preview, then Project GenieGenie 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

Interactive Worlds Beyond Video Clips

Genie 4 points to a different way of generating digital environments: rather than producing a finished clip, a model attempts to respond as a user explores and changes a scene. If the system becomes reliable and accessible, that could give game teams and filmmakers a faster way to sketch environments or test ideas. DeepMind says early partners have given feedback on film previsualization and game prototypes; these are reported uses, not evidence of a finished commercial product or measured productivity gains.

The technology also matters for research into simulation. DeepMind showed a robotics experiment in which a simulated Apptronik Apollo humanoid played checkers, with a separate model breaking the task into steps. The demonstration did not involve a real robot, according to the supplied reporting. Synthetic scenes may offer researchers ways to explore tasks, but the announcement does not establish that Genie 4 can replace physical testing or provide reliable training data for robots.

For audiences, generated scenes with synchronized sound may become more convincing. DeepMind says SynthID is embedded in outputs, but the announcement does not explain how people will identify that watermark in ordinary viewing or how well it will persist when content is edited or shared. That leaves open practical questions about provenance as synthetic video and audio become more realistic.

Products Worth Considering

OMRON Complete 2-in-1 Upper Arm Blood Pressure Monitor with EKG

OMRON Complete 2-in-1 Upper Arm Blood Pressure Monitor with EKG

Philips Sonicare 7300 Rechargeable Electric Toothbrush with 3 Brush Heads, Black

Philips Sonicare 7300 Rechargeable Electric Toothbrush with 3 Brush Heads, Black

As an affiliate, we earn on qualifying purchases.

From Genie Research to Project Access

Genie 4 follows earlier versions of DeepMind’s world-model work. The first Genie, introduced in February 2024, generated simple 2D worlds and was described in its research paper as an 11-billion-parameter model trained on unlabeled internet videos. Genie 2 followed in December 2024, generating 3D worlds from a single image and maintaining them for up to a minute, although real-time operation required a reduced-quality version.

DeepMind introduced Genie 3 in August 2025 as a real-time model running at 720p and 24 frames per second, with consistency lasting a few minutes and visual memory reaching back about a minute. Google later made Genie 3 available to paying U.S. subscribers through Project Genie, according to the source material. TechCrunch reported a 60-second limit per world, partly related to computing costs. Genie 4’s announcement does not provide a comparable public access limit or compute-cost breakdown.

Genie 4 is a model-generated environment, not a conventional game engine with a stored 3D map of every object. It predicts what comes next from the scene and user input. That approach can create responsive-looking worlds, but the lack of a fixed map is also relevant to DeepMind’s warning that details may drift.

““a few minutes of continuous interaction””

— Google DeepMind

Access, Duration and Reliability Questions

DeepMind has not announced a date for access, identified the early API partners or disclosed a Gemini API price. Google AI Ultra is described as the subscription route through Project Genie, but the source material does not confirm when that access will open or whether its availability will be geographically limited.

It is also unclear how Genie 4 performs outside the company’s selected demonstrations. No independent benchmark results, detailed error rates or direct comparison with other world models are provided in the announcement material. DeepMind’s “few minutes” description does not specify a standard session length, and the roughly two-minute figure reported by TechCrunch and Bloomberg is not in the company’s own account.

Claims about applications remain at different stages. Film previsualization and game prototyping are described as tester feedback reported by DeepMind, while the humanoid-robot scene was a research demonstration in simulation. The material does not establish that the model is ready for production use, or that its simulated physics will transfer reliably to real-world conditions.

The Next Test Is Wider Access

The next public milestone is the start of the promised limited rollout to Google AI Ultra subscribers through Project Genie and early Gemini API partners. Google has not said when that will happen, what the API will cost, or whether access will expand beyond those groups.

As testers gain access, useful evidence will include how long scenes remain coherent, how faithfully they follow text instructions, and whether generated sound stays aligned with what appears on screen. Independent testing and details about safeguards, watermark detection and practical costs would help clarify whether Genie 4 can move from selected demonstrations to dependable tools. Until those details emerge, the confirmed picture is a capable-looking research and product announcement with limited access and acknowledged drift.

Key Questions

What is Google Genie 4?

Genie 4 is a world model that generates a scene frame by frame and is designed to respond to user movement and instructions. DeepMind says it can also generate audio alongside the visuals.

Can the public try Genie 4 now?

No general public access has been announced. DeepMind says access is coming soon for trusted testers, including Google AI Ultra subscribers through Project Genie and early Gemini API partners. No date has been provided.

What are Genie 4’s main limitations?

DeepMind says objects and details can shift or drift over time and describes continuous interaction as lasting a few minutes. The company has not published benchmark results in the supplied material.

Is Genie 4 already being used to train real robots?

The cited robotics example was a simulation involving an Apptronik Apollo humanoid playing checkers; the supplied reporting says no real robot was involved. DeepMind’s announcement does not establish that Genie 4 is ready for real-world robot training.

Source: ThorstenMeyerAI.com

You May Also Like

Rebrandable client delivery dashboard for AI agencies

A new rebrandable client delivery dashboard for AI agencies is set for initial testing, aiming to improve client transparency and trust.

Critical Insights Into AI Sovereignty Testing Through The 24% Rule

An in-depth analysis of France’s SecNumCloud sovereignty rule and its impact on AI and cloud providers, highlighting the significance of ownership caps.

Simplify Workflows With These 14 AI Automation Software Solutions In 2026

Explore the 14 leading AI automation tools in 2026 that simplify workflows across industries, from agent builders to coding assistants, and learn why they matter.

VigilSAR: The Object That Isn’t Transmitting

VigilSAR uses SAR technology to identify objects that lack transponder signals, enhancing maritime awareness in all weather conditions.