Visko launches Orbis, a new real-time AI video model

Visko has launched Orbis, a generative video model built around real-time output at 4K resolution. The core distinction from existing video generation tools is that Orbis does not require a user to wait for a completed clip - it streams video continuously and responds to new prompts without stopping the feed. This makes it less like a video renderer and more like a live, steerable visual engine.

Most current AI video models operate in a batch fashion: a prompt goes in, processing happens, and a finished clip comes out after some delay. That workflow works well for pre-produced content but is poorly suited to applications that need responsiveness - live events, interactive installations, or real-time creative tools. Orbis is designed to close that gap by keeping the generation loop open and accepting input changes mid-stream.

The 4K resolution target is also significant. A number of real-time or near-real-time video models have traded image quality for speed, producing output at lower resolutions that limits their practical use in professional or display contexts. If Orbis can hold 4K quality while maintaining the real-time stream, it would address both constraints at once, though independent assessments of output quality at that resolution remain to be seen.

Visko is a relatively new name in the generative AI space, and Orbis appears to be its flagship model. The launch positions the company squarely in a growing area of competition around video generation, where larger players are also pushing toward faster and higher-quality output. How Orbis performs under real-world conditions - and how it will be made available to developers or end users - will likely determine how much traction it gains beyond the initial announcement.