Runway

Runway's latest research release, Solaris, is a real-time model that generates interactive software interfaces frame by frame, without any underlying code or predefined layout. Every frame is synthesized as you interact, allowing the interface to respond continuously to your actions. The company positions it as its first "Interface World Model," aimed at product designers, application builders, and teams training agents that need to operate across visual environments rather than fixed HTML or scripted screens. The motivation is that today's software is locked in during development. Every piece of software built today still requires a translation: the visual design must first be converted into an intermediate representation (e.g. code) before it can do anything. That intermediate representation limits what an interface can be, and how it responds to human and agent interaction. Solaris collapses that step by treating clicks and drags the way Runway's video models treat text prompts, producing whatever the next frame should be. It builds on the company's Gen-4.5 video generation model, adapted for interaction and sequential frame generation, and Runway highlights use cases from virtual storefronts where a user can drag a shirt onto their reference image, to interactive science demonstrations, to more adaptive training environments for computer-use agents that currently overfit to the specific layouts they were trained on. The launch matters now because it lands as Runway, led by co-founder and CEO Cristóbal Valenzuela, pushes deeper into world models beyond video. Solaris is Runway's first Interface World Model, and the company says it is working with key partners to explore what generated software can become , signaling that the same generative stack powering AI film tools is being aimed at the operating layer that both humans and agents will interact with next.