Visko has launched Orbis, a generative video model built around continuous interaction rather than one-off clips.
Instead of prompting, waiting for a finished video and starting again for every change, Orbis is designed to keep a generated world running while the user alters the scene. Visko describes support for text-to-world, image-to-world, voice control, live streaming and interactive story changes.
The company claims 4K output at 24 frames per second, immediate reactions to interventions and hour-scale visual coherence. Those are vendor-reported performance claims and have not yet been independently established at product level.
Orbis is publicly accessible through Visko’s site, although the freeform experience currently requires an account. The model is notable because it pushes generative video toward persistent, steerable environments that could eventually matter for games, simulation and interactive training as well as entertainment.