Details
- Runway unveils Solaris as its first Interface World Model, positioned as a new kind of operating system that generates software interfaces on the fly.
- Solaris produces interactive apps and websites frame by frame in real time, directly rendering the interface instead of compiling or executing traditional code.
- The system turns natural-language prompts and user actions into functioning user interfaces, with every frame synthesized as the user clicks, types, or drags.
- Runway reports that Solaris outperforms frontier large language models on generating new interfaces, citing stronger structural accuracy and better information retention in its experiments.
- Solaris builds on Runway’s Gen-4.5 video generation and broader world-model research, adapting frame-by-frame video prediction to treat mouse and keyboard input as conditioning signals.
- A separate language model handles reasoning and intent understanding, while Solaris focuses on continuously rendering the visual and interaction layer.
- The release is framed as research rather than a production OS; Runway is accepting early-access requests but has not detailed public release timing, pricing, or hardware requirements.
- Solaris extends Runway’s world-model work from physical environments into software UX, suggesting a path toward generated, persistent interfaces that evolve as users interact.
- By removing the need for explicit UI code, Solaris targets designers, developers, and AI-native products that want rapid interface experimentation without conventional frontend engineering.
- The announcement hints at future exploration of richer interactions, stronger grounding and longer-lived experiences built on Interface World Models.
Impact
Solaris pushes Runway’s world-model strategy beyond video into software, signaling an emerging category where interfaces are continuously generated rather than statically coded. This could pressure traditional UI toolchains and prompt rivals in AI agents and app builders to invest in generative operating layers. If scaled, such systems may lower experimentation costs for digital products and shift interface design toward prompt-driven workflows.
