Virtual try-onManipulate products inside a living visual scene.
SOLARIS · INTERFACE WORLD MODEL
The interface is no longer built. It is generated.
Solaris creates visual, alive and open-ended interfaces in real time—generating every next frame from the way you interact.
Visual InterfaceLiving WorldsOpen-Ended
ModelRequired
Solaris IWM
Starting frameImage
＋ Drop a starting image or browse
Interaction promptOptional
Create a living product showroom. Let me rotate the object, change materials and explore the scene.
Interaction mode720p
Click & DragAmbientTransition
10 credits per interactive experience
Living InterfaceGenerated frame by frame
LIVE · 720PInteractive examples
One model. Infinite interfaces.
Solaris turns a visual starting point into an experience that changes with every click, drag and instruction.
Living interiorsExplore and transform spaces without fixed screens.
Interactive learningLearn by acting directly inside the generated world.
A new interface primitive
Visual interfaces. Living worlds.
Solaris removes the boundary between an interface and the world behind it.
Generate from a starting frame
Give Solaris an initial image or state. It turns that visual seed into an interface you can directly manipulate.
Interact without fixed screens
Click, drag and type naturally. Solaris responds by generating the next frame instead of loading a pre-coded page.
Keep the world alive
Objects, light and motion remain active over time, creating coherent sessions instead of isolated outputs.
Core capabilities
Built for interaction, not playback.
Entirely visual
The interface is generated as pixels. No fixed component library or predefined screen is required.
Alive by default
Motion, lighting, objects and state continue to evolve while the user interacts.
Open-ended
Go beyond a narrow flow. Solaris can interpret new actions and generate a fitting next state.
Designed for possibility
Experiences that software could not hold before.
Adaptive commerce
Try, rotate, restyle and compare inside one continuous scene.
Living spaces
Explore interiors that respond immediately to every visual decision.
Hands-on learning
Turn explanations into environments where people learn through action.
Agent training
Create open-ended visual environments for embodied reasoning and testing.
Model overview
What is an Interface World Model?
Solaris is a generative model for interfaces. Instead of selecting the next pre-built screen, it reasons about your action and renders the next visual state in real time.
The result is an interface that can be entirely visual, continuously alive and open-ended—bringing world-model behavior into everyday interaction.
IWMModel class
720pVisual quality
Real-timeResponse
Frame by frameGeneration
Click · drag · textInputs
ContinuousState
How Solaris works
Reason, render, respond.
A continuous generative loop turns intent into the next visible state.
Learn interaction
Solaris learns how visual worlds respond to clicks, drags, language and changing context.
Render in real time
Each response appears as the next frame, preserving the visual continuity of the session.
Reason and continue
The model uses the current state and your next action to keep the experience coherent and open.
Interface behavior study
Generative interaction feels more natural.
In the reported evaluation, people preferred Solaris interactions over coded interfaces on instruction following and natural behavior.
| Evaluation | Solaris | Coded interface | Preference gap |
|---|---|---|---|
| Follows the instruction | 61% | 24% | +37 pts |
| Behaves naturally | 71% | 21% | +50 pts |
| Interface construction | Generated in real time | Predefined screens | Open-ended |
Figures reflect the comparative study reported in the Solaris research introduction.
A world behind every action
Watch an interface come alive.
No scene cuts. No fixed page flow. Just a visual world responding as it is used.
Access Solaris
Start generating living interfaces.
Try the research experience with credits already connected to your account.
Free access
Explore Solaris and begin with the credits available on your account.
$0
- ✓ Google account access
- ✓ Interactive generation
- ✓ 720p visual output
Credit packs
Generate more Solaris experiences with flexible account credits.
10 credits
- ✓ Per interactive experience
- ✓ Click, drag and text input
- ✓ Continuous visual state
Build what is next
Explore the first steps toward interfaces that generate themselves.
Open-ended
- ✓ Living visual systems
- ✓ Agent-ready environments
- ✓ New interaction patterns
Frequently asked questions
Everything to know about Solaris.
What is Solaris?
Solaris is an Interface World Model: a generative model that creates interactive visual interfaces frame by frame in real time.
How is it different from a normal website or app?
Traditional interfaces are assembled from predefined screens and components. Solaris generates what you see and how it changes as you interact.
How do I interact with a Solaris experience?
Start from an image, then use clicks, drags and text instructions. The model interprets the action and renders the next visual state.
What output quality does Solaris support?
The current research preview focuses on real-time interaction with 720p visual quality.
What is Solaris best at?
It is designed for visual, alive and open-ended experiences including commerce, spatial exploration, interactive learning and agent simulation.
Does Solaris preserve state?
Solaris is designed for sustained interaction and long-session coherence, so a scene can evolve continuously rather than reset after every action.
How many credits does one generation use?
Starting an interactive Solaris experience uses 10 credits. Your balance is checked before entering the experience.
Where can I try it?
Use any Generate or Try Solaris button on this page. Sign in with Google, and you will be taken to the existing chat experience when credits are available.
The interface is the world
Ready to interact with what comes next?
Start with one frame. Solaris turns it into a world you can use.