Runway Solaris: The AI Model That Could Change How We Build Websites and Apps

What Is Runway Solaris?

Solaris is Runway's first Interface World Model, introduced on August 31, 2026.

In a traditional application, developers create an interface using technologies such as HTML, CSS, JavaScript, React, or other frameworks. Every button, animation, interaction, and workflow has to be programmed in advance.

Solaris takes a different approach. Instead of creating a fixed interface and then programming its behavior, Solaris generates the interface continuously and responds to user interactions in real time.

In simple terms: Traditional software is programmed first and then rendered. Solaris generates the interface and its responses together.

Runway says Solaris can generate each frame as the user interacts with the application, allowing the interface to continuously change based on clicks, drags, typing, and other actions.

When Was Solaris Released?

Runway announced Solaris on August 31, 2026.

However, Solaris is not yet a normal publicly available Runway model like a typical video-generation tool. Runway describes it as an early step toward a new class of interfaces and is working with partners toward a public launch.

Currently, users can request early access through Runway's official Solaris page: Request Solaris early access on Runway.

How Does Solaris Work?

The easiest way to understand Solaris is to compare it with a traditional website.

Traditional Website:
User ↓ Click a button ↓ JavaScript detects event ↓ Application logic runs ↓ HTML/CSS changes ↓ New interface appears
The developer has already defined what should happen.

Solaris:
User interaction ↓ Language model understands intent ↓ Solaris understands visual state ↓ Solaris generates next frame ↓ User sees result ↓ User interacts again ↓ Next frame generated

This process continues while the user interacts with the interface. Runway says Solaris treats user interactions such as clicks and drags as conditioning signals for generating subsequent frames.

Solaris Uses AI for Both Reasoning and Rendering

One of the most interesting parts of Solaris is that Runway separates reasoning from rendering:

• A language model determines what should happen next.
• Solaris determines how that change should visually appear.

For example, imagine an interactive furniture application where a user says: "Change the color of the couch."

The language model interprets the request and determines the intended change. Solaris then generates the visual result:

User request ↓ "Change the couch color" ↓ Language model ↓ Understands action ↓ Solaris ↓ Generates updated scene

This is different from a traditional application where a developer would have to write specific code for changing the couch color. Runway describes this architecture as separating reasoning from rendering: the language model determines how the application should evolve, while Solaris generates the visual state.

Why Is Solaris Different From Normal AI Video Models?

Traditional AI video models are primarily designed to create videos from a single prompt (Prompt ↓ AI generation ↓ Finished video).

Solaris is designed for continuous interaction (Starting scene ↓ User interaction ↓ Generate next frame ↓ User interaction ↓ Generate next frame...).

The important difference is continuous interaction: the user isn't simply watching an AI-generated video; the user is interacting with an AI-generated environment.

What Can Solaris Do?

Runway highlights three major characteristics of Solaris:

• 1. Solaris Is Visual: The interface can be created directly from a visual scene rather than being limited to traditional UI components. For example, in a virtual clothing store, instead of clicking 'Add to cart', a user could pick up a shirt, drag it onto themselves, and interact directly with the environment.
• 2. Solaris Is Alive: Traditional websites usually wait for an action before changing. Solaris is designed to continuously render the environment—objects can move, lighting can change, and the environment responds continuously.
• 3. Solaris Is Open-Ended: Traditional applications are limited by predefined workflows. Solaris aims to make interactions flexible, responding dynamically to whatever the user does.

How Could Solaris Change Websites?

Imagine a traditional shopping website with fixed navigation (Home → Products → Categories → Cart → Checkout) where every user receives essentially the same interface.

With an Interface World Model, the experience becomes dynamic. For instance, if a user states: "I want to buy a sofa for my small living room", the AI generates a personalized showroom where the user moves furniture around and the AI continuously updates the environment.

The website becomes less like a collection of pages and more like an interactive environment.

Solaris Could Also Change Online Learning

One of the most interesting potential applications is education. Imagine learning how a combustion engine works. Instead of reading a paragraph or watching a pre-recorded animation, an AI-generated interactive environment allows students to experiment with different materials and observe the results in real-time.

ENGINE ↓ PISTON ↓ Drag the piston ↓ Change pressure ↓ Observe reaction

The tutorial responds to the learner's actions instead of forcing everyone through the same sequence. Runway specifically describes interactive demonstrations such as combustion experiments as a key use case.

Solaris and AI Agents

Solaris could also be important for AI agents. Today's computer-use agents generally interact with traditional websites and can become overly dependent on layout structures. If a website changes its layout, an agent may struggle.

Solaris introduces a different possibility: agents could train against interfaces that continuously change and may never have existed before, creating more adaptable computer-use agents.

What Technology Powers Solaris?

Solaris builds on Runway's Gen-4.5 video generation model, adapted to understand user interactions and generate frames at interactive speeds. Runway describes three important technical steps:

• Step 1 - Autoregressive frame generation: Instead of generating an entire video clip at once, Solaris generates frames sequentially (Frame 1 ↓ Frame 2 ↓ Frame 3...), where each frame depends on what happened before.
• Step 2 - Faster generation: Traditional video diffusion models require many denoising steps, which is too slow for real-time interaction. Runway distills that process into fewer steps for rapid frame generation.
• Step 3 - Training on its own outputs: Runway trained the model on its own generated outputs to maintain visual quality during longer interactions.

Why Hasn't This Been Done Before?

Building an interactive AI-generated interface presents three major engineering challenges:

• Speed: Delay over ~0.5 seconds breaks the feeling of interactivity.
• Consistency: Maintaining objects, layouts, text, visual identity, and scene state across long sessions is difficult for generative models.
• Cost: Real-time frame generation for every user action is significantly more expensive than serving static HTML/CSS/JS.

What Can't Solaris Do Yet?

Despite the excitement around Solaris, it is still an early technology with key limitations:

• Text: Generating stable, perfectly readable text in real-time video remains an open challenge.
• Trust: Ensuring generated outputs are grounded in verified real-world facts (crucial for commerce, finance, medical, etc.).
• Long Sessions: Preserving consistency over extended open-ended interactions.
• Accessibility: Interfacing with screen readers and accessibility APIs.

Will Solaris Replace React and JavaScript?

Not today.

Solaris is a new research direction rather than an immediate replacement for traditional web development. React, HTML, CSS, JavaScript, and backend APIs remain essential for reliable production applications.

A future application could potentially blend traditional code (for business logic, payments, data, and accessibility) with generative models (for dynamic, interactive visual experiences).

What Does Solaris Mean for the Future of Software?

Solaris imagines a future where software moves away from fixed application catalogs toward dynamic interfaces that generate around what the user wants to accomplish. Instead of asking "Which app should I open?", users may simply state "What do I want to do?", and the operating layer generates the interface for that exact task.

Final Thoughts

Runway Solaris represents a fascinating shift in AI—moving from generating static text, images, and video to generating the interactive interface itself. While significant engineering challenges remain, Solaris gives us an early glimpse of what could come after the traditional application interface.