Grok Imagine Video

What it does

Grok Imagine Video is xAI's video-editing model on Pipe2. You give it a video you already have plus a plain-language instruction, and it returns an edited video. Its one capability is video-to-video, so both the input and the output are video files. It does not generate a clip from a text prompt alone, and it does not take a still image as a starting frame. Edits are billed against short segments of output, with a six-second edit as the standard unit.

When to use it

• You already have footage and the change you want is easier to describe in a sentence than to specify with sliders or masks.
• You want to alter an existing clip (its look, its contents, its treatment) rather than generate a new scene from nothing, which this model cannot do.
• You are working in short segments: the billing unit is a six-second edit, so cost is easiest to reason about on clip-length work.
• You want the edit to stay anchored to your source video, since the source is a required input rather than an optional reference.

What to know before you run it

You must supply a source video. There is no text-to-video and no image-to-video path here, so if you have nothing to edit yet, generate the clip with another model first and bring the result to this one. Cost scales with the duration of the output. Pipe2 prices this model per second of edited video, so a longer edit costs more than a shorter one, and the six-second unit is what the price table is built around. Instructions carry the whole edit, so specific, concrete wording gives you more control than a broad request. Check the live price table on this page before running a long clip.

Also known as

You may see this model referred to as Grok Imagine Video 1.5, or by its slug grok-imagine-video-1.5. Both point at the same model listed here.