Runway Explores Streaming AI Video Generation

Runway is researching a way to generate AI video as a continuous live stream that users can steer in real time, rather than waiting for finished clips to render.

Runway is investigating a shift away from the standard AI video workflow of submitting a prompt and waiting for a finished clip. Instead, the company is exploring a streaming model where footage is generated and displayed continuously while users adjust prompts and direction on the fly, closer to interacting with a live feed than requesting a rendered file. The effort builds on Runway's GWM-1 world model, which produces video frame by frame rather than generating a whole clip at once. That frame-by-frame structure is what makes a live, steerable stream technically feasible, since the system can theoretically respond to new input as it keeps generating rather than locking in a fixed sequence upfront. According to The Decoder, a key focus of the research is cutting down the time it takes to produce the first frame of video, which is central to making real-time interaction feel responsive rather than laggy. Runway's ambitions for this approach reportedly extend past creative and entertainment applications. The company is also looking at how frame-by-frame, real-time video generation could serve fields like robotics and autonomous driving, where systems need to simulate or predict visual environments continuously rather than in isolated clips. While still framed as research rather than a shipped product, the direction signals where Runway sees the next phase of video generation heading: away from static outputs and toward dynamic, controllable streams that respond moment to moment to what a user or system asks of them.