Runway takes a swing at the operating system: Solaris generates interfaces frame by frame, in real time
Solaris, Runway's first Interface World Model, draws apps and sites with no code as you interact; the company says it beats frontier LLMs at generating new interfaces.
Runway, the video-generation pioneer, has cracked open a very different door: Solaris, the company's first Interface World Model. The idea starts with a question: what happens when an operating system generates apps and websites as you use them? Solaris does not run fixed applications; it renders the interface layer itself, in real time, frame by frame, with no code. Every frame is synthesized as you interact, so the interface reshapes instantly under your touch.
By Runway's measurement, Solaris outperforms frontier LLMs at generating new interfaces on two axes: structural similarity and information retention. The screens it produces both look more like real interface patterns and lose less of their information as the interaction unfolds. An early-access request form is open.
The team is candid about the open problems: stable, legible text remains one of the hardest challenges in video generation, and interfaces depend on text more than almost any domain; for trust, the starting frame can be composed from real product imagery, and conditioning generation on verified context (reference images, product data, documents) as a session unfolds is an active research focus, as is keeping visual and semantic coherence over long sessions.
The bigger picture: Runway recently shipped its real-time world model GWM-1, and Solaris is the screen-facing side of the same bet. The world-model race has jumped from physics simulation to the interface; tomorrow's operating system may not run apps at all; it may imagine them on the spot.
