Runway claims its GWM-1 “world models” can stay coherent for minutes at a time

Lastly, GWM Avatars combines generative video and speech in a unified model to produce human-like avatars that emote and move ...