Innovation & Tech

Genie 4 Opens for Eligible Google AI Ultra Users in Project Genie

Google’s Nov. 18 rollout gives eligible AI Ultra subscribers first access to Genie 4, including mid-session world edits and native audio, while SynthID marks all generated video and audio.

By Daniel Okafor

4 min read

Google Brings Genie 4 to Project Genie for Eligible AI Ultra Users
Google Brings Genie 4 to Project Genie for Eligible AI Ultra Usersterydanphiri / pixabay

What's News

  • November 18, 2026: Google announced a Genie 4 rollout for eligible Google AI Ultra subscribers through Project Genie.
  • October 27, 2026: Google DeepMind named Project Genie Ultra subscribers, early-access Gemini API partners and Google AI Studio partners as the next release groups.
  • Genie 3's per-generation limit increased from 60 seconds to three minutes.
  • January 29, 2026: The Genie 3-powered Project Genie prototype began rolling out to U.S. Google AI Ultra subscribers aged 18 and over.
  • Every Project Genie video and audio output includes SynthID; every model input and output receives safety screening.

Google announced on November 18, 2026 that eligible Google AI Ultra subscribers would receive first-time access to Genie 4 through Project Genie. The rollout places Genie 4 inside Google's browser-based world-building prototype, alongside a conversational creation agent and a community gallery.

Google says members of Google AI Pro can join a waitlist on the Project Genie site. The initial rollout excludes that group.

What changes during a Genie 4 session?

Google says Genie 4 lets users type prompts while navigating to change weather, add a character or object, move the camera, or redirect events, applying each adjustment without restarting the simulation.

Its native audio runs with the visuals. Google says that audio includes ambient sound, music and character dialogue. The dialogue reacts to user inputs and events inside the simulation.

How does Gemini keep the simulation coherent?

Google DeepMind describes Genie 4 as “natively multimodal.” Gemini Omni supplies a real-world reference for how simulations look and behave. Other Gemini models provide reasoning for coherence and continuity.

According to Google DeepMind, Gemini 3.8 Flash can serve as an in-loop guiding agent. It observes the scene, splits goals into steps, checks incoming frames for progress, and changes its plan when the world or user input shifts.

Google DeepMind presents this coupling as a way to model multi-step events and test how AI systems navigate environments before deployment in the real world begins. Google lists training environments for robots and agents, plus new interactive experiences, as intended uses.

How does creation move into the gallery?

Google says the Project Genie agent handles the initial creation through natural-language conversation, coordinating Google's multimodal models to produce the world's opening visuals and a custom background score suited to its intended mood.

Google says Gemini can watch the screen during exploration, suggest context-aware next actions through on-screen chips, and trigger events automatically. The gallery adds a publishing and reuse layer to that workflow.

Google says users can ask the agent to alter a world's structure, soundtrack or rules, then publish their version for others to explore and remix. Google's examples include Living Memory, which converts personal photographs into explorable spaces. The Life of Ants, created with filmmaker Anand Gandhi, presents a colony from an ant's perspective.

What can Genie 3 still do?

Google positions Genie 4 as the higher-fidelity option for visuals, physics and native audio, while Genie 3 remains available for longer runs and worlds seeded from 360-degree panorama images.

Google raised Genie 3's per-generation limit to three minutes from 60 seconds. Users can link multiple worlds into longer sequences. Natural-language prompts can also trigger real-time events.

How did Genie 4 reach Project Genie?

Google DeepMind's October 27, 2026 research preview introduced Genie 4 to trusted testers and named Ultra subscribers in Project Genie, Gemini API partners with early access, and Google AI Studio partners as the next release groups.

The November 18 announcement delivers the Project Genie portion of that plan. The original Project Genie prototype, powered by Genie 3, began rolling out on January 29, 2026. Its initial audience comprised U.S. Google AI Ultra subscribers aged 18 and over. Its workflow covered sketching worlds from text and images, navigating environments as Google generated them, and remixing existing creations.

What remains unresolved?

Google says every Project Genie video and audio output includes SynthID, its “imperceptible watermark” for identifying AI-generated content. Google also screens every model input and output for safety. Google describes both Genie models as frontier research systems.

Google DeepMind lists camera latency, intricate physical interactions, real-time audio precision and long-term scene permanence among the unresolved challenges facing both frontier research systems. In Genie 4, visual details and object permanence can drift during longer sessions, after complex edits or while users navigate. Legible text and spatial accuracy also remain active research areas.

For users, the next test is whether Genie 4 can preserve visual detail, object permanence and audio precision through longer sessions as they navigate, edit or extend a simulated world. Camera lag, scene drift and audio imprecision still limit extended creative work and Google DeepMind's proposed training environments for robots and agents.

The rollout expands who can build with Genie 4. Durability will determine whether it becomes practical for extended creation and the proposed training of robots and agents.

Source: unite.ai

Share this article:

More from Daniel Okafor

Daniel Okafor

Show full bio

Correspondent covering money, funding and market shifts around entrepreneurship, startups and venture funding at Business Bearings.

2,360 articles

Related articles

« Previous articleNext article »