EN Grok Imagine is xAI’s video model and the cheapest way to make a clip in the bot, at two credits a second for 480p. It renders 480p, 720p and 1080p natively, runs from 6 to 30 seconds, and every clip comes back with sound whether you asked for it or not.
What it is good at
This is the model to think out loud with. At two credits a second a six second draft costs 12 credits, which is roughly a fifth of the smallest credit pack, so you can try a framing five or six times before committing the idea to something more expensive. Most people end up drafting here and re-running the version they liked on a heavier model.
The sound is the other reason to reach for it. Grok is the only model in the catalogue that always returns audio and never asks you about it, so a clip arrives finished rather than silent. Open the bot and it is the model a fresh draft starts on.
What it will not do
It takes one photo, not several, so multi-reference setups belong on Wan 3.0 or Seedance 2.5. It does not accept a reference video at all. And while the aspect ratio list is generous, including the portrait 2:3 and landscape 3:2 that the other models mostly skip, there is no 21:9 here.
Picking a resolution
Draft at 480p. The difference between 480p and 720p is two credits a second, which sounds small until you multiply it by a 30 second clip, where it is 60 credits against 120. Move up once the shot is right rather than while you are still deciding what the shot is. Every rate is on the pricing page.