Details
- Runway introduced Solaris on August 31, 2026, describing it as the first model in a new family called Interface World Models that generates interfaces frame by frame in real time.
- The system is part of Runway and builds on its Gen-4.5 video model and earlier GWM-1 world-model work, combining language-model reasoning with visual rendering.
- Solaris is designed to let clicks, drags, and typed instructions directly reshape the interface, eliminating the usual code translation layer between design and execution.
- Runway said its evaluation compared Solaris with a coded-interface baseline and with frontier multimodal models on screenshot reconstruction, and that users preferred Solaris in instruction-following and natural-behavior tests.
- The company also said Solaris is not yet broadly public and is being prepared with partners for launch, with current work focused on real-time interaction, long-session coherence, and 720p visual quality.
Impact
Solaris points to a broader shift toward generated interfaces and more dynamic AI agents that can operate in changing visual environments, rather than static app layouts. If the approach matures, it could influence product design, computer-use automation, and agent training by reducing reliance on fixed UI code and improving adaptability across sites and workflows.
