Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering

Google DeepMind put Genie 4 into research preview, and the jump from Genie 3 is substantial: more photorealistic worlds, generated audio including dialogue and music, mid-simulation prompts to change weather or camera angle or add people, and scenes that stay consistent for up to two minutes, double the prior limit.

It lands as the world model market consolidates. TechCrunch reports AMD is acquiring World Labs, Amazon is in talks to buy Decart, and Rivian bought Third Dimension for self-driving simulation. Runway shipped its own playable model with audio a couple of months ago. Whoever owns these systems will shape how games, training, and robot testing are built.

I'm most interested in the hybrid design. Gemini now acts as a director that plans steps and checks frames while Genie draws the world. In a robotics experiment, a Gemini model broke tasks into steps that Genie 4 simulated in real time. Testing robot behavior in simulation before it touches the physical world is one of the more clearly useful things AI can do, and cheaper, richer simulation could make real robots safer.

The trusted testers' pitch is also refreshingly grounded: prototyping films, storyboarding, rapid ideation. A tool that keeps a creative team in flow while they sketch a scene is a good use of generative AI. It assists the work rather than pretending to finish it.

Google is candid about the limits. Objects drift over longer runs, audio accuracy is unsolved, and in-world text can be illegible. Two minutes of consistency is progress, not permanence. That honesty matters, because the downside of world models is the gap between a convincing simulation and a correct one. If robot training or driver testing leans on worlds whose physics quietly drift, failures could get baked in rather than caught.

The other downside is cultural. One speculative example Google floated is stepping into a historical painting and chatting with its subjects. That could be a great teaching tool. At scale it could also blur the line between history and plausible invention, with synthetic dialogue from real people delivered in convincing audio. I'm watching for provenance and labeling to arrive alongside the realism.

Our read: Genie 4 is a meaningful step and Google is right to keep it in preview. The consolidation wave tells us the industry thinks world models are infrastructure. Infrastructure needs verification standards before it needs more photorealism.

Genie 4 is available to trusted testers now. Google says it will roll out in the coming weeks to Google AI Ultra subscribers in Project Genie and to early-access partners of the Gemini API and Google AI Studio. Google also offers Veo for video generation, which has found an audience among filmmakers.