September 10, 2026, 02:00

Grok video for businesses: What a usable clip really costs

AI-generated from the sources listed, with automated source checks.

xAI's video model can generate up to 15 seconds of video from text or images. According to a third party, the per-second price is only part of the bill, because discarded clips and review also cost money.

xAI has an API that lets developers and companies make short videos with AI. An API is a technical gateway that lets other programs use the model. According to xAI's own documentation, the grok-imagine-video-1.5 model is used to create video from text or still images, while grok-imagine-video is used to edit an existing video.

What it can do

A clip can be at most 15 seconds long. Length, aspect ratio and resolution can be set. The API supports up to 14 reference images and 3 voice references, which control how the video looks and sounds. You can also lock frames at the start, the end and the middle of the video, and extend a video from its last frame.

Videos are generated asynchronously. You send a request, get a request ID and poll repeatedly to check whether the video is finished. In other words, you don't wait in front of the screen while the clip is being made.

The price per second

xAI says pricing is charged per second and depends on duration and resolution. The specific rates are listed on xAI's pricing page. The documentation itself doesn't give them.

According to the provider WaveSpeed, a third party, as of September 10, 2026, grok-imagine-video-1.5 costs USD 0.08 per second at 480p, USD 0.14 at 720p and USD 0.25 at 1080p. Duration can be 1-15 seconds, and 480p is the default. An image input costs USD 0.01 per image, text input has no separate media price, and video is delivered with audio by default at no extra charge. WaveSpeed also says the preset voice is currently free.

The finished clip costs more

WaveSpeed points out that the per-second price doesn't cover clips you discard. You should therefore budget for how many clips end up being approved.

WaveSpeed's own worked example covers a 10-second shot made from an image. Three drafts at 480p cost USD 2.43, and one final version at 1080p costs USD 2.51. That adds up to USD 4.94 in generation costs. Add USD 6 for human review and USD 0.20 for processing, and you land at USD 11.14 per approved clip. The review figure is based on WaveSpeed's own assumption, and the company has a commercial interest in the area.

Finally, a practical detail: according to the sources, xAI credits can neither be refunded nor transferred between organizations.

This means the price of a video is not just the per-second price. For those considering using AI video in their work, time spent on drafts and review is an important part of the calculation. The prices come from a third party and may have changed.