Google Deepmind's Genie 4 adds real-time sound, Gemini planning

Genie 4 builds on the video model Gemini Omni for realistic rendering, while other Gemini models plan the simulation. For example, Gemini 3.8 Flash watches the scene continuously, breaks a goal into steps, and checks whether each step worked using new frames. Deepmind is also testing Gemini 4 Argon to tie Genie to its robotics research, Gemini Robotics; in one demo, Argon breaks a checkers task into steps that Genie 4 uses to simulate Apptronik's humanoid robot Apollo playing the game, with no real robot involved.

Deepmind calls Genie 4 an early, experimental model. The world stays consistent over short sessions, but visual details can shift and objects don't always persist in longer simulations or after several complex prompt changes. The company says it's still researching complex physical interactions, accurate real-time sound, and readable text.

Looking ahead, Deepmind sees pairing Gemini with Genie as an early step toward hybrid world models that combine the reasoning of language and code models with pixel-level simulation. Once they mature, world models could serve as training environments where robots and AI agents learn from simulated experience rather than just from data, according to the company.