The next generation of models won't just generate images - they'll understand worlds, motion, interaction, and action. We've been building toward this for a while.

Visual intelligence is becoming real-time. @stephenbtl spoke about where we're headed at @aiDotEngineer:

Video