Google DeepMind releases Genie 4 world model in research preview to trusted testers
Article timeline
The articles in this story. Times are ET.
- MarkTechPostNewsPickGoogle DeepMind releases Genie 4, a real-time world model with native audio
AIGoogle DeepMind has released Genie 4 as a research preview for trusted testers, generating explorable worlds from text and image prompts in real time. It produces ambient audio, music and character dialogue alongside the visuals, and a Gemini model runs inside the simulation loop as a planner. DeepMind discloses no parameters, resolution, frame rate or benchmarks, and lists drift and short sessions of a few minutes as open limitations.
Ethan MollickXEthan Mollick demos Genie 4 generating a real-time, walkable Victorian tea sceneAIEthan Mollick says Genie 4 generated a Victorian high tea scene with a mouse character, where a porcupine walked by and the mouse grew wings and flew to a teacup. He reports the environment runs in real time and he can walk around freely in it, though the output is glitchy.
indigoXGoogle launches Genie 4, built on Gemini Omni, for realistic real-time world simulationAIGoogle's Genie 4 has launched and is built on Gemini Omni. The author says it renders realistic detail in spaces and specific themes, with advanced physics, fully real-time control, ambient sound, music, and interactive dialogue. It plans multi-step events on Gemini reasoning to keep scenes coherent as they evolve.
KhazixXGoogle releases Genie 4, a world model with real-time control and native audioAIGoogle has released Genie 4, its next-generation world model, which the author says adds stronger photorealism, better physics, real-time control by keyboard or natural language, and native synchronized audio. The article also describes a Gemini-based guiding agent that plans and checks steps while Genie 4 generates the world. The author contrasts this video-generation approach with World Labs' 3D Gaussian splatting approach, noting that the former offers dynamic generation but weaker consistency and higher inference cost.
- SiliconANGLE · AINewsPickGoogle previews Genie 4 world model for generating interactive 3D environments
AIGoogle previewed Genie 4, an AI model that generates realistic three-dimensional environments from natural language descriptions. Users can edit the simulation with follow-up prompts or keyboard camera controls, and Google is testing it with a limited number of early adopters before making it available through Google AI Ultra, the Gemini API, and Google AI Studio.
Demis HassabisOfficialPickGoogle DeepMind's Genie 4 world model adds real-time physics, reasoning, and audioAIDemis Hassabis announces Genie 4, an interactive world model that combines real-time physics, reasoning, and audio. The quoted Google DeepMind post says users can explore dynamic environments in real time and see how the scene responds, sounds, and how people in it move, react, and speak.
- The DecoderNewsGoogle DeepMind introduces Genie 4, a real-time world model with sound and dialogue
AIGoogle DeepMind has introduced Genie 4, the successor to Genie 3, which generates interactive worlds frame by frame with directly generated sound and talking characters. Gemini models plan the simulation in the background, and the company says Genie 4 is currently available only to select testers as a research preview. DeepMind calls it an early, experimental model, noting that visual details can shift and objects may not persist in longer sessions.
Philipp SchmidOfficialPickPhilipp Schmid announces Genie 4 with native audio and live prompt editsAIPhilipp Schmid announces Genie 4, a real-time multimodal world model that renders playable environments frame by frame from live keyboard controls and text prompts. The post says the update adds native audio, characters that move and talk, sharper physics, and live prompt edits. Access starts with a waitlist, with Google AI Ultra subscribers first.
Logan KilpatrickXGoogle previews Genie 4, a frontier world model coming to the APIAIGoogle's Logan Kilpatrick says an early look at Genie 4, Google's new frontier world model, is out. Genie 4 is coming soon to the API for the first time.
Rohan PaulXGoogle DeepMind's Genie 4 adds real-time native audio and in-simulation editingAIGoogle DeepMind announces Genie 4, a real-time world model that generates ambient sound, music and live dialogue from characters who react to users and in-world events. Users can rewrite a running simulation by typing, changing weather, moving the camera or adding people, while Gemini fits each edit to the scene. Access starts with trusted testers, followed by Google AI Ultra subscribers and early Gemini API partners.
- Ethan MollickXEthan Mollick demos Genie 4 generating interactive game-like environments in real time
AIEthan Mollick says Genie 4 generates game-like environments as he types, producing a neogothic brutalist city with robots and horses in about a minute. He reports it added dragons and a friendly otter NPC on request, all in real time, while noting it is not flawless.
- TechCrunch · AINewsGoogle DeepMind releases Genie 4 world model in research preview with audio and real-time steering
AIGoogle DeepMind released Genie 4 in research preview, adding real-time audio, mid-simulation prompt edits, and scenes that stay consistent for up to two minutes. Gemini plans steps and checks frames while Genie 4 generates the visuals. The model still has gaps: details can drift over long simulations, and audio accuracy and in-world text legibility are unresolved. Access starts with trusted testers, then Google AI Ultra subscribers in Project Genie and early Gemini API and Google AI Studio partners.
Andrew CurranXIt's finally here, Genie 4 has been released in research preview, and will be available to Ultra subscribers soon. Some of my friends wer...AIIt's finally here, Genie 4 has been released in research preview, and will be available to Ultra subscribers soon. Some of my friends were in the closed test, so maybe they can finally talk about it now. I'm really looking forward to this.
🚨 AI News | TestingCatalogXGoogle DeepMind announces Genie 4 world model with native audio and dynamic promptingAIGoogle DeepMind announces Genie 4, a world model that users can prompt to change the world dynamically and generate with native audio. It is rolling out to Trusted Testers as a research preview, with Google AI Ultra subscribers and early access partners to follow via the Gemini API and AI Studio.
Sundar PichaiOfficialPickGoogle announces Genie 4, a world model with real-time steering and native audioAIGoogle's Sundar Pichai announces Genie 4, which it calls its most realistic and interactive world model. The model lets users steer environments in real time, generates audio matched to the scene, and uses Gemini reasoning for multi-step events. It is rolling out as a research preview to trusted testers, with access coming to Google AI Ultra subscribers in Project Genie and early-access partners via the Gemini API and AI Studio.
Google AIOfficialPickGoogle introduces Genie 4, a world model steered with natural languageAIGoogle AI announces Genie 4 from Google DeepMind, which it calls its most realistic and interactive world model yet. Users can steer the world, weather, camera and added elements with natural language in real time. Gemini acts as the brain that translates requests into actions and plans events, and Genie 4 adds ambient audio and real-time interactive dialogue.
GoogleOfficialGoogle DeepMind introduces Genie 4, an interactive world modelAIGoogle DeepMind announces Genie 4, its newest interactive world model. The post introduces the model without giving further technical details such as parameter count, benchmarks, or availability.
Google DeepMindOfficialGoogle DeepMind unveils Genie 4, a real-time interactive world modelAIGoogle DeepMind announces Genie 4, which it calls its most realistic and interactive world model yet. The post says users can explore dynamic environments in real time, with scenes that respond, sound, and show people who move, react, and speak.
- Google DeepMindOfficialPickGoogle DeepMind releases Genie 4, a real-time world model with native audio
AIGoogle DeepMind says Genie 4 is a new world model that generates interactive, photorealistic simulations with native audio, including ambient sound and dialogue. Gemini models guide the simulation to keep scenes coherent, and the company is rolling Genie 4 out first as a research preview to trusted testers. Access will follow for Google AI Ultra subscribers in Project Genie and early partners in the Gemini API and Google AI Studio.
- Google DeepMind · YouTubeOfficialPickGoogle DeepMind unveils Genie 4, a real-time world simulation model
AIGoogle DeepMind announces Genie 4, a world model built on its Gemini models that lets users move through dynamic worlds in real time and steer scenes with natural language. Genie 4 is rolling out as a research preview to trusted testers, and the company says it is coming soon to Google AI Ultra subscribers in Project Genie and to early access partners of the Gemini API and Google AI Studio.
- Ars Technica · AINewsGoogle DeepMind unveils Genie 4 world model with more realistic physics and humans
AIGoogle DeepMind has unveiled Genie 4, an updated world model that Google says improves on Genie 3 in almost every way. World models generate interactive virtual environments frame by frame from a text description, without pre-rendered assets. Google says the model adds audio and more realistic people and physics, and that updated Gemini models help it simulate what users request.
- Google DeepMind · The KeywordOfficialPickGoogle DeepMind announces Genie 4, a world model for real-time interactive simulation
AIwith native audio. The model builds on the Gemini family, using Gemini Omni for real-world appearance and behavior and other Gemini models for reasoning. Users can steer scenes with natural language, but the model is currently limited to trusted testers, with broader access coming to Google AI Ultra subscribers and early access partners.
- The Verge · AINewsGoogle announces Genie 4 world model with longer, more realistic interactive experiences
AIGoogle announced Genie 4, a world model for generating interactive experiences, as a research preview for trusted testers. The company says it creates more realistic worlds, updates experiences in real time from natural language prompts, generates audio, and uses Gemini 3.8 Flash for simulations. Clips run about two minutes, up from one minute for Genie 3, though Google warns that details and object permanence can shift over longer simulations.