Google releases Genie 4 to AI Ultra subscribers with real-time prompt editing and native audio
Google is rolling out Genie 4, which it calls its most realistic and interactive world model yet, to eligible Google AI Ultra subscribers in Project Genie. The update adds real-time native audio, mid-session text prompting, a world-building agent and a community gallery. Genie 3 also gains sessions of up to 3 minutes.
Google began rolling out Genie 4 on November 18, 2026 to eligible Google AI Ultra subscribers through Project Genie. Google describes it as its most realistic and interactive world model yet. Google AI Pro members can join a waitlist on the Project Genie website.
The announcement was written by Diego Rivas (Google DeepMind) and Pavan Kumar (Google Labs). Project Genie is described as an experimental research prototype, introduced earlier this year, for generating and exploring interactive worlds in the browser.
What Genie 4 adds
- Mid-session prompting: Users can type a new prompt while navigating a world to edit it in real time, for example shifting the weather, adding a character or redirecting the action.
- Native audio: Genie 4 is the first Genie model to pair its visuals with real-time native audio that adapts as the scene unfolds, from ambient soundscapes to character dialogue.
- Fidelity: Google says Genie 4 delivers its highest-fidelity visuals, physics and native audio of the Genie models.
Google has not yet disclosed Genie 4's parameter count, benchmark scores, context window or API pricing.
Genie 3 updates
Genie 3 stays in Project Genie's model picker for longer runs and panorama-seeded spaces. Google lists three updates:
- Extended sessions and world chaining: Up to 3 minutes per generation, up from 60 seconds, plus the ability to link multiple worlds into longer journeys.
- 360-degree panorama image grounding: New worlds can be seeded from 360-degree panorama images.
- Real-time in-world events: Natural-language prompts can change the environment while the user explores.
Project Genie agent and gallery
The new Project Genie agent lets users build worlds through conversation rather than a single text or image prompt. According to Google, the agent orchestrates its suite of multimodal models to produce initial visuals that set the world's look, along with custom background scores matched to the mood.
Gemini models can also join a session. Google says they observe what is on screen, surface suggestion chips for where to go next, and trigger in-world events automatically.
The new Project Genie gallery combines curated worlds with community creations. Users can publish their own worlds, and others can explore and remix them, including changes to world structure, soundtrack or rules. Featured examples include:
- "Living Memory," which turns personal photos into explorable spaces.
- "Dino Field Guide," built on 360-degree panoramas.
- "Rock Paper Scissors," with Gemini as a real-time referee.
- "Villa Giallo," a gothic-manor mystery made with creative technologist Will aka BLVCKL!GHT using Genie 4.
- "The Life of Ants," made with filmmaker Anand Gandhi, which uses Genie 4's photorealism to put the player in a colony from an ant's perspective.
Safety and limitations
All video and audio generated in Project Genie carries SynthID watermarking, and all model inputs and outputs go through safety screenings. Google says Genie 3 and Genie 4 remain world models in the frontier research space. It lists camera latency, intricate physical interactions, real-time audio precision and long-term scene permanence as challenges it is actively working on. A Genie 4 model research overview covers limitations in more detail.
What this means
Genie 4 is available only inside Project Genie in the browser, and only to eligible AI Ultra subscribers. Pro members are on a waitlist. Developers cannot yet access it directly, and no API or pricing details have been given.
The release pairs the world model with other components. The agent and Gemini models handle world setup, suggestions and in-world events, while the gallery adds publishing and remixing. Google keeps Genie 3 selectable because it handles longer runs (up to 3 minutes) and panorama-seeded worlds, so users choose between Genie 4's visuals, physics and native audio and Genie 3's session length and panorama grounding.
The limitations Google lists, including latency, physical interactions, audio precision and scene permanence, mark where the research is still unfinished.
Related Articles
Google adds "temporary threads" to AI Mode in Search, detaching chats from your account
Google is adding "temporary threads" to AI Mode in Search, a chat option that detaches the session from the user's account. The conversations won't resurface in search history or recommendations. The feature is rolling out globally on desktop and mobile across all supported languages.
Gemini Notebook now builds interactive websites from your notes with a "make me a site" prompt
Google's Gemini Notebook, formerly known as NotebookLM, can now turn a notebook of research into an interactive website from a text description. TechRadar tested it on King Arthur research and got a finished site in about five minutes, though the reviewer says claims need checking against sources.
Google makes Gemini side panel in Apps Script generally available, with existing admin controls applying
Google announced on November 16, 2026 that the Gemini side panel in Google Apps Script (GAS) is generally available. Users can create, edit and debug scripts with natural-language instructions. According to Google, scripts run within users' existing permissions and standard OAuth consent, so admin console visibility and control remain.
Gemini Skills now support sharing and live Google Drive file references, rolling out to all Workspace and personal accou
Google has added sharing and file attachments to Skills, the custom-instruction feature in the Gemini app. Files from Google Drive and Gemini Notebook update in real time, and each skill accepts up to 100MB of uploaded files. The rollout covers all Google Workspace and personal Google account users.
Comments
Loading...