Runway Solaris: Interfaces Generated in Real Time

On August 31, Runway introduced Solaris — the first model in a new family it calls Interface World Models. Instead of generating a video clip from a prompt, Solaris generates the interface itself, frame by frame, in real time: an app, a storefront, a dashboard that is continuously rendered as you interact with it. The model builds on Runway's Gen-4.5 video foundation, adapted to understand interaction and respond within the half-second threshold where interfaces stop feeling interactive.

It's research with an early-access form, not a shipping product — but the direction is the story for ecommerce teams, because the demo reel is basically a catalog of shopping scenarios: a virtual clothing store where you drag a shirt onto your own photo to try it on, a furniture scene where you say "change the color of the couch" and it does, a storefront that adapts its products, layout, and recommendations to each visitor in real time.

Here's what's actually new, what's still vapor, and what it means for teams whose job is selling products with video.

What Solaris actually is

The core claim is removing the intermediate layer. Today, a visual design has to be translated into code — an intermediate representation — before it can respond to a user. That translation is a lossy compression: behavior has to be anticipated and implemented ahead of time, and visual fidelity gets sacrificed for responsiveness.

Solaris collapses that. A single world model generates every frame and every response to user input — the entire frame is the interface. Three engineering challenges had to fall for it to work, and Runway's framing of them is worth reading even if you never touch the model:

1. Speed. Interactions stop feeling interactive around half a second of delay. Video diffusion models take seconds or minutes per clip — acceptable for content creation, too slow for an interface. Solaris generates frames sequentially, each depending only on what came before, cheaply enough to keep up with the user.
2. Coherence. An interface has to stay consistent across a whole session — text, layout, object identity — the exact things generated video has historically struggled to maintain. Small errors compound the longer generation runs.
3. Cost. Rendering every frame is still more expensive than serving a built-once page. Runway says the work that made Solaris real-time also made it orders of magnitude cheaper to run than a standard video diffusion model (vendor-reported) — with the cost curve still improving.

For ad teams, that third point is the familiar one: the render layer keeps getting cheaper and more general. The same trajectory that commoditized video production — which is why AI video ads now cost single-digit dollars — is pointed at interfaces next.

Why this matters beyond the demo reel

Runway's own post sketches where this goes, and the phrasing matters for anyone who sells online:

• The app stops being the unit you interact with. Instead of a fixed catalog of apps — one for shopping, one for news, one for reservations — "what you need simply shows up, customized to you."
• A storefront stops being a fixed layout every visitor sees. It becomes a generated environment that preserves brand identity while adapting to each individual — products, layouts, colors, materials, and recommendations reshaping around intent in real time.
• Tutorials stop replaying the same sequence. They render the next step in your context and recover naturally when you go off script.

Read those three against how ecommerce actually works today. Product pages are still HTML that a developer builds; A/B tests are still discrete variants someone designed in advance; personalization is still a rules engine on top of a fixed template. Solaris is the bet that all of that becomes generated output — and that the generation layer is what adapts, not the layout.

The practical signal for a media buyer is not "rebuild your store in a world model." It's that the unit of creative is about to get wider than the video file. If interfaces can be generated in real time, the ad experience stops at "video ends, landing page begins" — the landing page itself can become part of the creative, personalized per visitor, generated on the fly. That's a version of creative testing that doesn't exist on today's platforms, and the teams with a fast, structured creative pipeline are the ones positioned to exploit it when it arrives.

The honest caveats

Solaris is a research announcement with an early-access request form, not a product on a pricing page. Runway explicitly says it's "an early step toward a new operating layer" and that it's working with partners on a public launch. There are no published prices, no public API, no confirmed timeline — treat every demo video as curated, and don't re-plan your 2026 Q4 around it.

The second caveat is the one we keep coming back to with every platform-native AI release: generation is not the same as performance. A model that can render a storefront on the fly still doesn't know your offer, your margin, or your audience's objections. The script, the hook, and the product story remain the part of the work that determines whether the creative converts — the render is increasingly a commodity. That's exactly the pattern we flagged when Google put Gemini Omni in Asset Studio: production commoditizes, differentiation moves up the stack.

What to do now

1. Keep the video pipeline fast and structured. If interactive, personalized ad experiences are the next format, the teams that test them first will be the ones whose creative process is already a repeatable pipeline — not a manual production cycle. A URL-to-ad pipeline that turns a product page into a finished UGC-style ad in minutes is the same muscle, aimed at today's formats. That's what Prizmad's workflow is built for: product URL in, finished ad out, variant after variant.
2. Watch the cost curve, not the demo. Runway's "orders of magnitude cheaper" claim is vendor-reported and unquantified. What matters is the direction: every generation layer that gets cheaper widens the space of what's worth testing. When interface generation has a real price, run the same math you run on video ad costs — per-experience cost, per-conversion value, not per-render awe.
3. Own the offer layer. Whoever ends up rendering your storefront — Solaris, a competitor, or a platform's built-in tool — the brand message, the hook, and the offer structure still have to be written by someone who understands the product and the buyer. That's the layer that doesn't commoditize, and it's the layer worth investing in now.

Solaris is a research step, and the responsible read is "direction confirmed, timeline unknown." But the direction — interfaces that render themselves, personalized per visitor, at a falling cost — is the same direction video already went. The teams that built their creative muscle on fast, verifiable iteration are the ones that will be ready when the format lands.