Key Features

The model produces audio alongside interactive video.
The preview reports continuous 720p video at 24 frames per second.
Generated audio is described at 48,000 Hz.
WorldPrompt combines persistent scene context with timestamped actions and camera input.
Text actions can address individual subjects or the scene itself.
The event format supports actions with overlapping time intervals.
Per-frame translation and rotation specify camera movement.
The demonstrated world continues from new input without a preset session length.

Its WorldPrompt representation separates persistent world context from a timestamped stream of events. Scene descriptions, subjects, physical conventions, and an initial frame establish context; overlapping text actions and per-frame camera inputs control evolution. The project reports 720p video at 24 frames per second with 48kHz audio.


The research points toward interactive entertainment, virtual characters, embodied-agent environments, and creative design interfaces. Unlike a fixed clip, a session has no preset length in the demonstrated formulation. The announcement is a research preview and does not establish a generally available free or paid production endpoint.

Get more likes & reach the top of search results by adding this button on your site!

Embed button preview - Light theme
Embed button preview - Dark theme
TurboType Banner

Subscribe to the AI Search Newsletter

Get top updates in AI to your inbox every weekend. It's free!