Google launches Genie 4 world model with real-time audio and scene editing

The model maintains consistency for up to two minutes, doubles Genie 3's duration, and integrates with Gemini for task planning and simulation. It supports interactive prompts and real-time audio generation but still faces issues with object drift, audio fidelity, and in-world text legibility.

Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.