xAI's video model for short clips with sound. Write a prompt or start from an image — 1 to 15 seconds, up to 720p.
Direct the frame. Choose the engine.
Grok Imagine is xAI's video generation model. Give it a written prompt or a still image, and it renders a short clip with synchronized sound — clips run from 1 to 15 seconds at up to 720p.
Two things set Grok Imagine apart from most video models. The first is audio: it generates sound in the same pass as the picture, so footsteps, engine noise and spoken lines land on the clip without a separate audio step. The second is pace. Grok Imagine is built for short, punchy shots — a product beat, a reaction, a single camera move — rather than long cinematic sequences, which makes it a natural fit for social posts and ad cuts.
It accepts both text and a starting image, so you can start from a prompt alone or animate a photo you already have. On Genreel it sits in the same model picker as Seedance, Veo, Kling and Sora — one prompt box, one credit balance, no second account. Run the same prompt on Grok Imagine and on one other model, and you can see the difference in motion, sound and cost before committing credits.
Everything happens in the generator at the top of this page. No install, no timeline, no plugins.
Step 1
Open the model list under the prompt box and select Grok Imagine. The parameters panel updates to what this model offers: 480p to 720p, 1 to 15 seconds, and audio included. The generate button shows the credit cost before you start.
Step 2
Describe one shot: who or what is on screen, the single action that happens, and how the camera behaves. Because Grok Imagine produces sound with the picture, name the audio in the same sentence — "waves breaking and gulls calling" is a valid part of the prompt.
Step 3
Choose a duration between 1 and 15 seconds and a resolution up to 720p. Shorter clips cost fewer credits. If the shot is going to a Reel or Short, switch the aspect ratio to 9:16 before generating rather than cropping afterwards.
Step 4
Press generate and wait while the clip renders. When it is ready, preview it in the browser, download the MP4, or change one line of the prompt and run it again. Every result is saved in your account. Failed renders return their credits automatically.
These are the settings available for Grok Imagine in the generator today. The model picker is the final word — it always shows the live options and the credit cost for the length you have selected.
Grok Imagine is priced as a premium model. A 5-second clip costs 160 credits, and that price already includes the sound. Audio is generated in the same pass as the picture, so what you see and what you hear come from the same render.
Because the credit cost follows the length you choose, set the duration to the shot you actually need. A longer clip costs more than 160 credits; a shorter one costs less. On the Lite yearly plan, 24,000 credits are included and the displayed rate is $0.00445 per credit — about $0.71 for that 5-second clip. The exact figure depends on the plan and the length, and the pricing page lists the current rate.
Sound uses the same 1 to 15 second window, so a spoken line has to finish inside the clip. Resolution tops out at 720p, so treat Grok Imagine as the fast, social-ready option and move to Kling 3.0 or Veo 3.1 when a deliverable needs 1080p or 4K.
1 to 15 seconds. The default is 5 seconds.
480p or 720p. Resolution does not change the credit cost.
Generated with the video. There is no separate audio charge and no switch to turn it off.
A text prompt, or a text prompt plus one starting image. Grok Imagine does not take a last frame.
160 credits for a 5-second clip. The price scales with length and is shown on the generate button before you start.
All four of these models can make a video from a prompt. They disagree about what matters most: sound, resolution, motion or clip length. Here is how Grok Imagine compares.
Grok Imagine and Sora 2 both generate audio as part of the clip. Kling 3.0 and Veo 3.1 treat sound as a switch. If a shot needs a spoken line, a door slam or traffic noise baked in, Grok Imagine and Sora 2 save you a step.
Grok Imagine runs 1 to 15 seconds. Veo 3.1 is 4, 6 or 8 seconds. Sora 2 is 4, 8 or 12. Kling 3.0 goes up to 15 seconds and Seedance 2.5 up to 30, so a scene that needs room to play out is not Grok Imagine's job.
Grok Imagine tops out at 720p. Kling 3.0 reaches 1080p with a 4K option, and Veo 3.1 offers 720p or 1080p. A deliverable that has to hold up on a large screen should move to one of those two.
Kling 3.0 is the strongest on human movement and detailed close-ups. Veo 3.1 is the strongest on camera work and physics. Grok Imagine is built for fast, expressive shots that read well on a phone with the sound on.
A 5-second Grok Imagine clip is 160 credits, and longer clips cost more. Veo 3.1 Fast and Wan 2.2 Fast are the cheaper ways to test an idea. Draft on Wan 2.2 Fast until the motion reads correctly, then run the same prompt on Grok Imagine for the version with sound.
Grok Imagine clips are short, so the prompt has to earn its words. Five habits that hold up in practice:
This is the one place the model rewards you for something other models ignore. Name the sound where it happens: "a match striking, then low wind" tells the model what the clip should sound like as well as look like.
"She lifts the lid and steam escapes" is a shot. "She cooks dinner" is a scene, and 15 seconds is not enough for it. One action that finishes inside the clip reads better than three that do not.
Locked-off, slow push-in, side track, orbit. "Cinematic camera" tells the model nothing it can act on. On a short clip, one deliberate move fills the runtime better than a busy one.
Where it comes from, and what it touches. Low sun from the left, neon spill on wet asphalt — light sets mood faster than a list of style adjectives.
If the subject drifts, fix the subject line. If the pacing feels wrong, change the duration instead of rewriting the prompt. Rewriting everything means you cannot tell what helped.
A pattern that works on this model and on the others here: "[Framing] of [subject] in [place]. [One action with an ending]. The camera [one move]. [Light]. [Sound]."
Read the AI video prompt guideGrok Imagine is at its best on short shots where the sound belongs in the picture. Use these four briefs as a starting point in the generator above — write the shot, including what it should sound like.
Name the city, one gesture, and the ambient layer: cart wheels, a shout, traffic behind. The call and the street can land in one pass, which is why this shot belongs on a model with sound.
One action with a hard ending. Fix the camera low and to the side, and let the board's clatter carry the last second of the clip.
An atmosphere shot at close range, where a short duration helps. Six seconds is enough for water to gather and run.
A locked-off camera around a small object, with room tone in the prompt. Slow enough that a still image would carry the same subject.
New accounts receive 40 free credits after sign-up. That pays for one Wan 2.2 Fast clip at 480p for 5 seconds, and also for one 1-second Grok Imagine clip — both cost 40 credits. It does not pay for the default 5-second Grok Imagine clip, which is 160 credits. The generator charges the full clip before it starts, so 40 credits is not a partial payment toward a longer one.
Keep testing on Wan 2.2 Fast, where 40 credits buys a complete clip, and switch to Grok Imagine when the shot needs sound. Or buy a credit pack and spend it on Grok Imagine directly — a credit pack does not require a subscription. Every plan shows the cost before you generate, leaves the MP4 without a watermark, and returns credits automatically if a render fails.
Added after sign-up. Enough for one Wan 2.2 Fast clip, or one 1-second Grok Imagine clip. Not enough for the default 5-second clip.
Genreel does not stamp a logo on the MP4, including clips made with Grok Imagine.
Payment details are only needed if you buy credits. The price is shown before you generate.
What Grok Imagine does in the generator, and what a clip costs.
Grok Imagine is xAI's video generation model, available in Genreel's model picker. It turns a written prompt or a starting image into a clip of 1 to 15 seconds at up to 720p, with sound generated alongside the picture.
Yes. Grok Imagine turns a written prompt into a video, and it can animate a still image you upload. In Genreel it appears in the model list for both text to video and image to video.
Not for the default clip. New accounts receive 40 free credits, which covers one Wan 2.2 Fast clip at 480p for 5 seconds, or one 1-second Grok Imagine clip. A 5-second Grok Imagine clip costs 160 credits, and longer clips cost more. The full cost is shown and charged before you generate.
Between 1 and 15 seconds, set in the parameters panel. Sound is generated within the same window, so keep any spoken line short enough to finish inside the clip. Credits scale with the length you choose.
Yes. Audio is generated together with the video, not added as a separate step, and it is included in the credit cost. That is the main reason to choose Grok Imagine over the silent models in the picker.
480p or 720p. For 1080p or 4K, switch to Kling 3.0 or Veo 3.1 in the same picker.
Open the generator on this page, select Grok Imagine from the model list under the prompt box, write your shot and press generate. No install and no separate account. Sign-up takes seconds, and the first 40 credits are added automatically.
Pick the model, describe the shot and the sound, and download the MP4. A 5-second clip is 160 credits.
