Odyssey has opened a free preview of its AI system named Odyssey-3, which generates interactive 3D environments directly in the browser.

Users can move through these scenes in real time and observe environments that respond to their inputs and navigation.

How the system operates

The model interprets textual descriptions and constructs a navigable environment, assembling visual elements and behavior rules for runtime interaction.

Rendering and interaction logic run inside the browser so scenes update on demand as users change viewpoint or interact with objects.

Difference from video generation

Unlike video models producing linear, pre-rendered clips, Odyssey-3 yields stateful scenes that allow viewpoint changes and ongoing interaction.

This makes the output suitable for exploration and testing, rather than for producing single, fixed-motion sequences for consumption.

Guides, examples and ready prompts

The original material includes a step-by-step usage guide and three illustrative scenarios with ready prompts to reproduce and adapt in experiments.

• Indoor walkthrough scenario for testing object interaction and navigable detail in confined spaces.
• Open-area exploration scenario demonstrating continuous camera movement and environmental transitions.
• Interactive puzzle scenario that showcases state changes triggered by user actions within the scene.

The preview is offered at no cost and the accompanying guide provides practical prompts and setup advice for immediate trial and evaluation.