Genie 4 Is Now Available: Google Opens Its Interactive World Model to AI Ultra Subscribers in Project Genie

Google rolled out a new Project Genie experience to eligible AI Ultra subscribers with Genie 4, its most realistic interactive world model, plus a Project Genie agent, longer Genie 3 sessions and a community gallery for remixing worlds.

Christopher Ort
By Christopher Ort · 3 min read
Share
A person with a controller in front of a screen showing a misty forest.
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. What's in it
  5. Why it matters
  6. What's still open
  7. Sources
  8. i10x Perspective

Google has opened Genie 4, its most realistic interactive world model so far, to paying users. On 18 November 2026 Google DeepMind and Google Labs began rolling out a new Project Genie experience to eligible Google AI Ultra subscribers, with Genie 4, an agent that helps build worlds, upgrades to Genie 3 and a community gallery for sharing and remixing worlds.

i10X Quick Take

Quick Take

  1. Who gets it: eligible Google AI Ultra subscribers from today; Google AI Pro members can join a waitlist.
  2. Genie 4: edit a running world with new prompts mid-session, with real-time native audio for the first time in Project Genie.
  3. Genie 3 upgrades: sessions of up to 3 minutes per generation (up from 60 seconds), world chaining, 360-degree panorama grounding and real-time in-world events.
  4. Project Genie agent: turns a conversation into a world, using Google’s multimodal models for the look and a matching background score.
  5. Gallery: curated and community worlds that others can explore and remix.

In plain English

Genie does not make a video you watch. It makes a small world you can move through, like a video game generated on the fly. With this update you can type “make it rain” or “add a dragon” while you are inside the world, hear sound that fits the scene, and publish what you made so others can play with it.

What happened

In a blog post, Diego Rivas of Google DeepMind and Pavan Kumar of Google Labs said early testers wanted more control over what happens inside worlds and easier ways to build and share them. The new Project Genie experience “brings together our most advanced world models alongside our Gemini models,” they wrote, letting people generate, step inside and shape worlds from the browser.

Genie 4 was first presented in October as a research preview for trusted testers; our earlier report covers the Genie 4 model and its limits. This is the first time Ultra subscribers can use it directly.

What’s in it

  • Mid-session prompting: shift the weather, drop in a character or redirect the action while navigating.
  • Native audio: ambient soundscapes to character dialogue, adapting as the scene unfolds.
  • Genie 3 in the model picker: for longer runs or worlds seeded from panorama images.
  • Gemini in the session: it observes what is on screen, suggests where to go next or triggers events automatically.
  • Creator showcases: “Villa Giallo” with creative technologist Will aka BLVCKL!GHT and “The Life of Ants” with filmmaker Anand Gandhi.
  • Safety: all generated video and audio carries SynthID watermarking, and inputs and outputs go through safety screening.

Why it matters

For creators, world models move from demo clips to a tool with remixing and publishing, closer to a game engine than a video generator.

For developers and game studios, prompt-driven worlds with sound could speed up prototyping and previsualisation.

For Google, Project Genie becomes another reason to pay for its top AI subscription tier.

What’s still open

  • Google says both Genie models remain in the frontier research space, with open challenges in camera latency, complex physical interactions, real-time audio precision and long-term scene permanence.
  • When Google AI Pro members or other countries get access.
  • Whether developers get Genie 4 through the Gemini API beyond early-access partners.

Sources

  • Google, “Genie 4 is now available,” blog post by Diego Rivas and Pavan Kumar, 18 November 2026
  • Google DeepMind, Genie 4 model research overview

i10x Perspective

Video generators fight over the prettiest ten seconds. Genie competes on something different: worlds that react. Putting it behind the Ultra paywall with a gallery is a test of whether interactive AI media can find a real audience. More from Google and our Video & Image section.

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Timeline

What happened before

  1. Release

    Gemini 4 Argon

    1.05M context · $2 / $10 per 1M tokens

    Google's most capable model for complex coding, agentic, math and science tasks.

    ai.google.dev (source, opens in a new tab)
  2. Release

    Gemini 3.8 Flash

    1.05M context · $0.75 / $3.75 per 1M tokens

    ai.google.dev (source, opens in a new tab)
  3. Release

    Gemini 3.7 Flash

    1.05M context · $0.75 / $3.75 per 1M tokens · Deprecated 8 Oct 2026 (API routes to Gemini 3.8 Flash)

    ai.google.dev (source, opens in a new tab)
  4. Release

    Gemini 3.6 Flash & 3.5 Flash-Lite

    1.05M context · $0.75 / $3.75 per 1M tokens

    ai.google.dev (source, opens in a new tab)
  5. Partnership

    Google to pay SpaceX $920M a month for Colossus compute (~$30B through 2029)

    With Google · Compute · $30B

    Per SpaceX's SEC filing: about 110,000 NVIDIA GPUs from October 2026 to June 2029, ramping through September at a reduced fee; either side may terminate after Dec 31, 2026 with 90 days' notice (CNBC, Bloomberg).

    cnbc.com (source, opens in a new tab)
Full Google timeline ›

Editor's picks

  1. GAO Report: ChatGPT, Claude, Gemini and Grok Produced Viable Attack Plans After Safeguards Were Bypassed
  2. ResolveAI SRE-1: A 27B Model Trained to Find the Root Cause of Production Incidents
  3. Reson8 Rumble-2-Flash: Open-Weight 0.5B Speech Recognition Model Covers 40+ Languages
  4. Pika Enact Lets Creators Direct AI Video Actors and Cameras in Real Time
  5. Google Stitch Loop: An Always-On Product Agent That Fixes Security, Speed and Accessibility via Pull Requests