Grok Imagine Video: write the sound into the prompt (6s, 10s, 15s, 720p, 7cr)

Grok Imagine Video has native audio. A silent prompt still gets a soundtrack you did not choose.

Grok Imagine Video is filed as text-to-video. requires_image is false. audio is true. Durations: 6s, 10s, 15s, 20s, 30s. Max output: 720p. Quality chips include 480p, 720p, and 1080p; the catalog ceiling on this row is 720p. Aspects: 16:9, 3:2, 4:3, 1:1, 3:4, 2:3, 9:16. Display price: 7 credits. The description string says "Grok image-to-video generation"; the categories and the image requirement say you can start from text. Either way, the mix is on.

Native audio is not optional

Native audio means the clip comes back with a soundtrack the model wrote. Room tone, music, Foley, a line of dialogue you never specified. A prompt that only describes picture is still a mix brief. You just left the mix to the model.

Write the sound: who speaks, what they say, whether the bed is quiet, whether the room is a street or a booth. If you needed silence, say silence. If you needed a line, write the line. The 7-credit generate will produce audio either way.

The door is the AI video generator. The ranked list of rows that carry a mix is best model with audio. Best text-to-video model is the category list. Grok Imagine Video sits on both because the category is text-to-video and audio is true.

Six seconds through thirty, 720p

Duration chips go 6s, 10s, 15s, 20s, 30s. The title's 6 / 10 / 15 are the short end; 20s and 30s are on the same row. A silent-looking 30s prompt is thirty seconds of leftover score. Write the cue to the length you picked.

720p is the max output resolution. Do not prompt "4K cinematic" and expect the row to sprout a 4K option. Match the delivery to 720p or plan an upscale as a later edit-video pass.

You do not have to upload a still. If you have a locked plate and you wanted image-to-video, confirm you are on that job — this catalog row is text-to-video with no image requirement. The Grok roster also has Grok TTS for a voice file when the mix should not come from the picture model.

Seven credits is a full mix, not a silent preview

7 credits is cheap next to silent 1080p rows that bill 60+. That is the trap. Teams treat it as a disposable picture test and then ship the accidental soundtrack. The accidental soundtrack is the generate. If the bed fights the VO you planned to add, you are mixing against a 7-credit choice you did not make.

Styles listed: Cinematic, Documentary, Abstract. Those are picture registers. They do not mute the audio flag. Documentary still gets a mix.

Write the cue or inherit one

The test: play the take with the sound off, then on. If the on version contains a line, a riser, or a room you cannot use, the prompt was incomplete. Put the sound in the next prompt. Do not "fix it in captions." Captions will transcribe the line you did not write.

If the job is a silent product turn, this may be the wrong row. Silent image-to-video exists elsewhere. If the job is a 6s–30s scene with diegetic sound at 720p for 7 credits, this is the row — and the prompt includes the soundtrack.

FAQ

Does Grok Imagine Video always return sound?

audio is true. Treat native audio as on. A picture-only prompt still gets a soundtrack you did not choose. Write the cue, or write silence.

What lengths can I generate?

6s, 10s, 15s, 20s, and 30s. Max output 720p. 7 credits display. No image required. Category is text-to-video.

Can I use this as a silent preview and add VO later?

You can mute in the editor. The file you paid for still has a mix, and that mix may already contain speech. Prompt the sound you want, or pick a silent row.

Is this the Grok speech model?

No. This is the video row with native audio. Grok TTS is the voice-file row on the same Grok roster. Use TTS when you need a track without a picture.