Quick Answer
Google is rolling out Genie 4 through an updated Project Genie experience for eligible Google AI Ultra subscribers. The new model adds real-time native audio, mid-session text edits, and a conversational agent for creating interactive worlds. Project Genie remains an early research prototype for adults, so users should treat its generated worlds as experimental rather than finished production tools.
Key Takeaways
- Google began rolling out Genie 4 on November 18, 2026.
- Genie 4 adds adaptive ambient audio and character dialogue to generated worlds.
- Users can change weather, add characters, or redirect action with prompts during a session.
- Genie 3 remains available and now supports generations up to 3 minutes.
- Project Genie is available to eligible Google AI Ultra subscribers age 18 and older.
What is Genie 4 in Project Genie?
Genie 4 is Google’s new Project Genie model for creating and exploring interactive generated worlds. Google began rolling out Genie 4 on November 18, 2026, through an updated Project Genie experience for eligible Google AI Ultra subscribers, as described in Google’s rollout announcement. The release moves Project Genie beyond a fixed generated clip by allowing a user to interact with a world as it unfolds.
Genie 4 combines generated visuals with real-time native audio, including ambient soundscapes and character dialogue that adapt to the current scene. Google describes this as the first Project Genie model with that pairing. The practical value is that a world can respond through both visuals and sound, although Google still classifies Project Genie as an early research prototype rather than a general-purpose game development platform.
Project Genie also includes an agent that builds interactive worlds through natural-language conversation. The agent can use Google’s multimodal models to create visuals and custom background scores. Users who are comparing generated world tools with more conventional video workflows should keep the distinction clear: interactive worlds are designed for exploration and modification, while finished video often requires editing, sequencing, and export controls available in AI video editing workflows.
Who can use Genie 4 right now?
Genie 4 is available through Project Genie for eligible Google AI Ultra subscribers who are age 18 or older. Google’s Project Genie availability page states that the product remains an early research prototype, while Google AI Pro members can join a waitlist. The rollout status matters because a Google subscription alone does not establish that every account can use the updated experience immediately.
Google has not positioned Project Genie as a broadly available consumer service for every Google account. The access limit gives Google room to test a complex generative system with a narrower group of users while the product is still evolving. Users who do not see Project Genie in their account should confirm their plan and eligibility instead of assuming a missing feature reflects a device problem.
Google AI Ultra access also places Genie 4 alongside other higher-tier Google AI releases, including recent Gemini model updates. The practical implication is that Genie 4 is part of Google’s premium AI product strategy, not a standalone app with separately published access terms in the information Google provided. Readers following Gemini’s AI Pro and Ultra releases should expect eligibility and rollout rules to remain important when Google introduces experimental features.
How does Genie 4 let users change a world during a session?
Genie 4 lets users alter an active generated world with a text prompt during a session. Google says a user can change the weather, add characters, or redirect the action while the world is running. The feature matters because the prompt is not limited to the start of creation, which gives users a way to revise the experience without beginning an entirely new world.
Genie 4’s live editing model changes the role of the prompt from a one-time instruction into an ongoing control method. A user can begin with one environment and then request a different condition or event as the session develops. That flexibility may be useful for prototyping interactive scenes, educational concepts, and creative experiments, but Google has not presented Genie 4 as a replacement for conventional game engines or production pipelines.
Genie 4’s conversational agent provides another route to creation by turning natural-language requests into interactive worlds. The agent can generate visuals and custom background scores using Google’s multimodal models. The most sensible approach is to use clear, specific prompts that identify the setting, intended action, and desired change, then review the generated result before publishing or sharing it with others.
| Project Genie capability | What Google says it does | What it means for users |
|---|---|---|
| Live world editing | Changes weather, characters, or action through mid-session prompts | Users can revise an active experience instead of restarting it |
| Native audio | Adds adaptive ambient soundscapes and character dialogue | Generated scenes can respond through sound as well as visuals |
| Conversational agent | Creates worlds through natural-language conversation | Users can describe an interactive concept without a technical authoring interface |
| Community gallery | Lets users publish, explore, and remix worlds | Users can share creations and build from other published experiences |
What does real-time native audio add to Genie 4?
Genie 4 adds real-time native audio that changes as a generated scene unfolds. Google says the audio includes ambient soundscapes and character dialogue that adapt to the world, a capability outlined in Google DeepMind’s Genie 4 research overview. The addition matters because sound can make an interactive environment easier to understand and more responsive to a user’s actions.
Native audio means Project Genie generates sound as part of the world experience rather than presenting a silent visual environment. An environmental change such as rain can therefore affect the atmosphere of a session through audio as well as imagery. Google has not provided consumer-facing technical performance measurements for audio quality, latency, or consistency, so users should evaluate the feature as an experimental capability rather than assume uniform results across every prompt.
Genie 4’s audio feature may be especially relevant for creators who want to test concepts where setting and mood are important. Character dialogue can also make a generated experience feel more interactive, but users should review any generated speech before sharing it publicly. Creators who need stable recurring people or fictional figures may also need dedicated tools focused on consistent AI-generated characters, because Project Genie’s announced features focus on interactive worlds rather than character continuity guarantees.
What happens to Genie 3 after the Genie 4 rollout?
Genie 3 remains available in the Project Genie model picker after the Genie 4 rollout. Google expanded Genie 3’s session limit to up to 3 minutes per generation, increasing the prior 60-second limit. The longer limit gives users a reason to continue using the earlier model when they need a longer single generated session rather than Genie 4’s newly announced audio and editing capabilities.
Genie 3 also gained 360-degree panorama image grounding, linked-world sequences, and in-world natural-language events. Panorama image grounding allows a generated experience to use a 360-degree panorama image as a reference point, while linked-world sequences connect worlds into a sequence. In-world natural-language events allow changes within a world through language-based instructions.
Google’s decision to retain Genie 3 gives users a model choice instead of requiring an immediate move to Genie 4. The practical comparison is straightforward: Genie 4 emphasizes real-time native audio and active world modification, while Genie 3 offers its own updated set of world-building features and a longer stated session limit. Users should select the model based on the specific experience they want to generate, then test the result before sharing it.
How does the Project Genie community gallery work?
The Project Genie community gallery lets users publish worlds for other people to explore and remix. Google highlighted “Living Memory,” “Dino Field Guide,” and “Rock Paper Scissors” as example experiences in the new gallery. The gallery gives Project Genie a sharing layer, which can help users discover different ways the system can create interactive worlds.
Remixing allows a user to build on a published world rather than create every experience from a blank prompt. That can make experimentation more accessible, particularly for users who want to study how an idea has been structured. At the same time, users should review a remixed world carefully before publishing because generated visuals, audio, and prompts can produce results that differ from the creator’s intent.
Project Genie’s gallery also changes the privacy calculation for people using the service. A world intended for publication is different from a private experiment, so users should avoid placing personal information, sensitive images, or confidential ideas into prompts meant for sharing. Google’s safety screening and watermarking apply to generated materials, but responsible publishing still requires the user to consider what the prompt and finished world reveal.
How does Google handle safety and watermarking in Genie 4?
Google says all video and audio generated in Project Genie carries SynthID watermarking, while inputs and outputs undergo safety screening. The safeguards apply to Genie 4 and the broader Project Genie experience. Watermarking matters because it provides a marker for generated media, while screening is intended to evaluate material entering and leaving the system.
Project Genie’s safety measures do not remove the need for careful user judgment. Users should not enter personal, financial, health, or confidential workplace information into prompts, especially when they may later publish a world to the community gallery. The practical response is to keep prompts focused on fictional, public, or non-sensitive material and to inspect generated output before sharing it with another person.
Project Genie is limited to eligible users age 18 and older, which is an important access boundary for an early research prototype. Adults responsible for younger users should not treat the product as an unrestricted creative tool for children. Users who believe a generated world contains harmful or inappropriate material should stop sharing it and use Google’s available reporting or support channels rather than attempting to distribute or remix the content further.
Should Google AI Ultra subscribers try Genie 4?
Google AI Ultra subscribers who want to experiment with interactive AI worlds should try Genie 4, particularly if real-time audio and live text-based changes are useful to their project. Genie 4 offers a distinct set of capabilities: adaptive sound, character dialogue, conversational world creation, and mid-session editing. Those features make the product more flexible than a one-time visual generation workflow.
Project Genie is still best treated as an experimental research product rather than a reliable production environment. Google’s published information establishes the feature set, but it does not promise that every world will meet a professional standard or behave consistently across prompts. Users creating commercial media, educational materials, or public experiences should keep an independent review process and avoid relying on a single generated output.
For most eligible subscribers, the practical approach is to begin with a small fictional prompt, test one or two live changes, and compare Genie 4 with Genie 3 when session length is the priority. Users who need traditional editing controls can continue working with established creative software after generation. Google’s AI tools are expanding across creative products, including AI features in Adobe and Google workflows, but each tool serves a different part of the creation process.
FAQ
Is Genie 4 available to all Google users?
Genie 4 is not available to all Google users. Project Genie is rolling out to eligible Google AI Ultra subscribers age 18 and older, while Google AI Pro members can join a waitlist.
Can Genie 4 change a world after generation starts?
Genie 4 can change an active world after generation starts through text prompts. Google says users can alter weather, add characters, or redirect action during a session.
Does Genie 4 generate sound as well as visuals?
Genie 4 generates real-time native audio alongside visuals. Google says the feature includes adaptive ambient soundscapes and character dialogue that change as the scene unfolds.
Is Genie 3 still available in Project Genie?
Genie 3 remains available in the Project Genie model picker. Google increased Genie 3’s session limit from 60 seconds to up to 3 minutes per generation and added several world-building features.
Does Project Genie watermark generated media?
Project Genie watermarks generated video and audio with SynthID, according to Google. Google also says Project Genie screens inputs and outputs for safety, but users should still avoid entering sensitive information into prompts.
