Google’s new world model generates interactive, explorable environments with sound, more realistic worlds and characters, and scenes you can reshape live with natural language — launching October 27, 2026 as a research preview for trusted testers.
2026-10-27 · by Moe Ameen
On October 27, 2026, Google DeepMind announced Genie 4, its new world model for real-time interactive simulation, as a research preview available to a small group of trusted testers. A world model is different from a text-to-video generator like Sora or Veo: instead of rendering a fixed clip, it generates an interactive, explorable environment on the fly — you move the camera, shape the scene, and it responds in real time.
The headline advance is that Genie 4 is natively multimodal. Google describes it as its first real-time world model to generate audio natively, producing ambient noise, music, and interactive dialogue alongside the visuals rather than as a separate pass. It is built on Gemini: Google says Gemini Omni anchors each simulation in real-world appearance and behavior, while other Gemini models handle the reasoning that keeps a scene coherent and continuous.
You can steer a running simulation with natural language — moving the camera, changing the weather, or introducing a new person or object mid-scene. Google points to more realistic worlds and more humanlike characters, and multiple reports describe Genie 4 holding a consistent scene for up to about two minutes, roughly double Genie 3's limit.
At launch access is limited. Google said Genie 4 would roll out in the following weeks to Google AI Ultra subscribers inside Project Genie and to early-access partners through the Gemini API and Google AI Studio, with a waitlist. All generated video and audio carry SynthID watermarking. Google is candid about the limits, too: details and objects can distort over longer sessions or after multiple edits, and physics, audio accuracy, legible on-screen text, spatial accuracy, and long-term consistency remain open research areas. A Google representative said it is still early and that the end-to-end product experience is not yet defined.
You can't sign up for Genie 4 today — it's a trusted-tester research preview — but the moment is content, and that's where [Kompozy](/) turns a news cycle into a week of posts. 'What is a world model, and why does Google generating interactive worlds with sound matter?' is exactly the explainer your audience is asking for. Draft the take once and Kompozy fans it out: a [Persona Short](/glossary/persona-shorts) where your avatar breaks down Genie 4 in under a minute with captions burned in, a brand-exact [carousel](/glossary/hyperframes) walking through native audio, real-time steering, and the two-minute limit, a [blog post and newsletter](/glossary/output-buckets) for readers who want depth, and quote graphics pulled from your sharpest line — every piece held to one voice by your [Persona Brief](/glossary/persona-brief).
Then [Autopilot](/glossary/autopilot) schedules and publishes the set across the eight social platforms plus blog and email from a single review queue, so your explainer lands while the story is still hot instead of three days late. And when Genie 4 opens up and you can actually screen-record a steered simulation, the same pipeline turns that footage into B-roll behind your shorts. For the Gemini model powering it, see [Gemini Omni](/ai-tools/gemini-omni); for DeepMind's related world-model work, see our note on [Gemini 4 Argon](/news/gemini-4-argon).
Genie 4 is a world model from Google DeepMind, announced October 27, 2026, that generates real-time, interactive, explorable environments instead of fixed video clips. It is natively multimodal — it produces matching audio alongside the visuals — and you can steer a running scene with natural-language prompts.
Not yet for most people. At announcement it is a research preview limited to trusted testers. Google said it would roll out in the following weeks to Google AI Ultra subscribers inside Project Genie and to early-access partners through the Gemini API and Google AI Studio, with a waitlist. No standalone free consumer product has been confirmed.
The headline changes are native audio (Genie 4 is Google’s first real-time world model to generate sound with the visuals), more realistic worlds and characters, and real-time natural-language steering. Reporting also describes Genie 4 holding a scene consistent for up to about two minutes, roughly double Genie 3’s limit.
Not directly — it is a gated preview and its outputs are short, still-imperfect simulations. The practical move today is to cover the launch itself: explain what a world model is in your own voice and publish that across platforms with a content engine like Kompozy, then repurpose real Genie 4 footage once access opens.