September 1, 2026, (Inside AI) — Runway has introduced Solaris, an AI model that generates interactive app interfaces in real time. The company calls it the first in a new family of systems known as Interface World Models.
Solaris produces visual interfaces frame by frame, responding directly to user actions like clicks and drags. Instead of rendering a fixed app design, the model creates the interface as the user interacts with it.
This approach marks a departure from traditional software development, where designers create layouts and developers code them using HTML, CSS, and JavaScript. Solaris skips that pipeline entirely.
Why Solaris Could Redefine Software Interaction
Runway describes Solaris as an answer to a simple question. “Solaris starts with a question: what happens when an operating system generates apps and websites as you use them?” the company wrote in a blog post.
Read: ByteDance Unveils Seedance 2.5: China’s 30-Second AI Video Model
The model builds on Runway’s Gen-4.5 video model and its earlier GWM-1 world model. It generates one frame at a time. User actions become inputs that determine the next frame. A separate language model interprets intent and decides how the scene should change.
To enable real-time interaction, Runway optimized the video model to generate frames sequentially and reduced the number of processing steps per frame.
From Static Menus to Simulated Environments
Runway offers several examples of how Solaris could work. An online clothing store could become a virtual showroom. Users could drag a shirt onto an image of themselves to preview it.
A room-design app could let users tell the system to move a table or change a sofa’s color, and the interface would respond visually. A cooking app could let users drag ingredients into a bowl and watch the environment react.
Unlike AI chatbots that respond with text, images, or links, Solaris responds visually and continuously. If a user asks for an interactive demonstration of combustion, it could generate the visual scenario and let the user experiment inside it.
This could also help AI agents adapt to changing interfaces rather than memorizing fixed button locations.
Solaris remains experimental. Stable, legible text is a major challenge because the model is built on video-generation technology. Accuracy is another concern, as an AI-generated interface could look convincing while showing incorrect information.
Read: China’s Fysics AI Launches Physics-Based World Model, Bypassing US AI Road Maps
Runway said it is working on measures to keep Solaris grounded in verified product information and reference material. The system also needs to maintain consistency during long interactions and eventually work with accessibility tools such as screen readers.
The broader vision is an AI-powered operating system that generates the interface needed for each task on the go. Traditional apps are not disappearing yet. Solaris is still an early research system, and Runway is working with partners toward a public launch.
