Runway Unveils Solaris, 720p AI Model That Generates Apps in Real Time
Updated
Updated · runway.com · Aug 31
Runway Unveils Solaris, 720p AI Model That Generates Apps in Real Time
3 articles · Updated · runway.com · Aug 31
Summary
Runway introduced Solaris as its first “Interface World Model,” saying the system generates app and website interfaces frame by frame as users click, drag or type.
720p real-time rendering is the key claim: Solaris pairs a language model that interprets intent with a world model adapted from Gen-4.5 to stream interactive frames without translating designs into code first.
250 participants in a 30-example study preferred Solaris over a coded interface in 61% of instruction-following comparisons and 71% of natural-behavior comparisons, according to Runway.
30-interface reconstruction tests were used to argue that current multimodal models lose fidelity when rebuilding interfaces from screenshots, especially as visual complexity rises.
Runway said Solaris still struggles with stable text, long-session coherence, trust and accessibility integration, and is being prepared for public launch through early-access partners.
Will Runway's code-free model revolutionize software, or just create an unauditable nightmare for enterprise security?
When AI renders interfaces frame-by-frame, what happens to accessibility for users who rely on underlying code?
If apps become real-time video streams, how will we ever debug or secure the software of tomorrow?
Runway’s Solaris: Eliminating Code with Real-Time Generative Interfaces—A 61% User Preference Over Traditional UIs
Overview
Runway’s Solaris introduces a new era of interface design by eliminating the traditional code layer and generating applications and websites dynamically, frame by frame, in real time. By pairing a large language model with a video generation engine, Solaris learns the direct relationship between user actions and visual outcomes, allowing users to manipulate digital environments and change functionality on the fly using natural language. This approach bypasses static UI frameworks, enabling instant, adaptive interfaces for industries like e-commerce and education. However, it also brings challenges such as high computational demands, potential visual errors, accessibility gaps, and the need for new security and QA methods, all while showing superior fluid responsiveness in user studies compared to coded alternatives.