Runway pitches real-time video as the next interface
Runway argues that real-time, frame-by-frame video generation will replace text boxes as the main way people interact online, but its pitch rests on products it has already launched rather than new results.
Runway has published a research essay arguing that online interaction is moving past "text in a box" toward video generated live in response to what users say and do. The post announces no new model, benchmark or capability. It mainly restates the case for products Runway has already shipped.
The essay builds on GWM-1, which Runway calls its first "general world model" family and which launched last December. GWM-1 is an autoregressive model that generates video frame by frame in real time. It can be controlled with actions such as camera pose, speech or robot commands. It ships in three separately post-trained variants: Runway Characters for conversational avatars, GWM-Worlds for explorable environments and GWM-Robotics for robotic manipulation. Runway says it is working to bring all three under a single base model but gives no timeline.
Characters gets the most attention. It is an audio-driven model that turns a single reference image into an expressive conversational character, with eye movement, lip-sync and gestures. Developers can use it through an API at dev.runwayml.com, and anyone can try it in the Runway web app. Runway says BBC, R/GA, Silverside and Supersonik are building with it. It does not say what those companies are building.
The company lists games, education, workplace training and customer service as near-term uses. It points to a separate post on identity, consent and transparency for its approach to responsible deployment. Runway also repeats its earlier forecast that it will reach "human-scale world simulation" within five years. The post offers no figures on latency, cost or adoption.