DeepMind Unveils Genie 4: Real-Time Worlds With Sound
Google DeepMind has unveiled Genie 4, a model that builds interactive worlds in real time and, for the first time, adds live sound and character dialogue.
DeepMind’s artificial intelligence (AI) team has launched Genie 4, an environment generator. It’s the first version of the Genie series that supports real-time audio, including sound effects, music, and conversations. To remember its objective and adapt accordingly during runtime, it relies on Gemini 3.8 Flash. At present, it’s accessible to only a handful of people. But it will be made available to Google AI Ultra members via Project Genie and early access users of the Gemini API shortly.
The AI team at DeepMind acknowledges that scene descriptions and object memories tend to degrade over time. Hence, it’s just a proof of concept rather than a final product. Don’t expect to write your code on the CLI and see what happens. Instead, use it to train AI agents/robots in a controlled setting where they can make mistakes without causing any harm. Yeah, it’s got potential, provided DeepMind makes it public for all to use.
Source: deepmind.google
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles