On 31 August Runway introduced Solaris, the first model in a family it calls Interface World Models. Instead of an app with code underneath, every frame of the interface is generated the moment you touch it. For now it is research with early access, not a product.
- Solaris builds on the Gen-4.5 video model, reworked to respond to clicks and drags in real time, at 720p.
- In Runway's own test with 250 people and nearly 7,500 comparisons, people preferred Solaris over a coded interface in 61 per cent of cases for following the request.
- Runway itself lists the weaknesses: legible text, long sessions, accessibility for screen readers, and trust.
You drag a shirt off the rack onto your photo and you are wearing it. No button, no menu, no page anyone programmed in advance.
That is Runway's demo for Solaris. There is no code underneath. Every frame is drawn as you move the mouse.
Where the real test is
Not in the shirt, and not in the salad you assemble by dragging ingredients into a bowl. In the list of what Runway says it cannot do yet. Text comes first. An interface without stable legible text is a picture, not a tool: a price, a pay button, terms and conditions - exactly where a mistake costs money.
As a designer I see the other side too. The whole craft of interfaces rests on predictability: the button is where it was yesterday. Solaris goes the opposite way, because every time it may look different. For a shop window or a game that is a gift. For a banking app it is a headache.
The research is serious and the numbers come with honest caveats. As a working tool it is early. Runway itself points to the online storefront as one of the places for it, and that is where the first real test will be.