Today
Google launches two physical-AI models in limited preview
Google made two physical-AI releases on the same day: DeepMind's world model Genie 4 launched as a research preview, and the robotics reasoning model Gemini Robotics ER 2 entered public preview on Google Cloud. A world model predicts how an environment changes from one moment to the next, so robots or AI agents can practice inside it. Genie 4 generates interactive, realistic scenes in real time. DeepMind says Genie 4 is the first real-time world model to natively generate audio and character dialogue, and DeepMind showed Gemini 4 Argon paired with Genie 4 simulating a humanoid robot playing checkers. Genie 4 is open only to trusted testers for now. DeepMind's post acknowledges that object permanence drifts in long simulations, and the post gives no evaluation figures (Google DeepMind, 2026-10-27). Gemini Robotics ER 2 handles a robot's high-level planning and calls lower-level action models as tools. Google Cloud reports 91.3% accuracy at finding key-event frames; Google doesn't say which frames were tested or which models it was compared against (Google Cloud, 2026-10-27). How we checked →

Source and checks
This item ran in the brief of Oct 28, 2026, alongside 15 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.
Cite this item
SecondSource (2026). "Google launches two physical-AI models in limited preview." https://secondsource.io/en/product/2026-10-28-2/. Accessed access date.