Updated
What is the Grok Imagine 1.5 Video API?
The Grok Imagine 1.5 Video API is live on UnificAlly as xai/grok-imagine-1.5-video, from $0.04 per second of video. Grok Imagine 1.5 Video is the official API build of xAI's Grok Imagine 1.5 video model. Send a prompt alone for text-to-video, one start_image_url to animate a frame, or up to 14 image_urls to guide subject, style and scene. Add an end_image_url to pin the last frame, up to 4 timed keyframes, and up to 3 preset voice_ids for spoken lines. Videos run from 1 to 15 seconds at 480p, 720p, or native 1080p, with sound effects, ambience, and dialogue generated in the same pass.
It is the quality route of the family and the one with the most control. On the Arena image-to-video board it is the highest-ranked xAI video model and sits ahead of every Veo 3.1 variant. For NSFW output, style presets, saved characters or 30 second clips, use Grok Imagine 1.5 Fast Video, the grok.com build of the same model.
Grok Imagine 1.5 Video API specs
| Model ID | xai/grok-imagine-1.5-video |
| Endpoint | POST https://api.unifically.com/v1/tasks |
| Request types | Text-to-video, image-to-video from a start frame, reference-to-video |
| Duration | 1 to 15 seconds |
| Resolution | 480p, 720p, 1080p. Reference-to-video renders up to 720p |
| Aspect ratio | auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 21:9, 5:2 |
| Audio | Soundtrack on by default, up to 3 preset voices, audio: false for a silent clip |
| Reference inputs | One start frame or 1 to 14 reference images, plus an end frame and up to 4 keyframes |
| Rate at 480p | $0.04/sec |
| Rate at 720p | $0.07/sec |
| Rate at 1080p | $0.125/sec |
Billing is per second of output, with no subscription. A start frame, an end frame, reference images and keyframe images each add a small per-image charge. The pricing page shows the live rates.
How to use the Grok Imagine 1.5 Video API
- Create an API key on the API keys page.
- Send a POST request to
/v1/taskswith the model ID and aninputobject. - Read the
task_idfrom the response, then check the task until it is finished, or set acallback_urlon the request and wait for the result.
curl -X POST https://api.unifically.com/v1/tasks -H "Content-Type: application/json" -H "Authorization: Bearer YOUR_API_KEY" -d '{
"model": "xai/grok-imagine-1.5-video",
"input": {
"prompt": "A lighthouse on a cliff at dawn, waves crashing below",
"aspect_ratio": "16:9",
"duration": 8,
"resolution": "720p"
}
}'
The response returns the task ID straight away:
{
"code": 200,
"success": true,
"data": {
"task_id": "abc123def456",
"status": "processing"
}
}
The finished task carries the video URL.
curl https://api.unifically.com/v1/tasks/{task_id} -H "Authorization: Bearer YOUR_API_KEY"
Grok Imagine 1.5 Video API parameters
| Parameter | Type | Default | What it does |
|---|---|---|---|
prompt | string | "" | Text prompt guiding the video. Required unless an image input is supplied |
aspect_ratio | string | auto | auto, 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 21:9, or 5:2. auto lets the model choose. Ignored for image-to-video, where the start frame governs |
duration | int | 6 | Video length in seconds, 1 to 15 |
resolution | string | 480p | 480p, 720p, or 1080p. Reference-to-video (image_urls) renders up to 720p |
start_image_url | string | none | Start frame to animate, as a public https:// URL. Cannot be combined with image_urls |
image_urls | string[] | none | 1 to 14 reference images for reference-to-video. Cannot be combined with start_image_url |
end_image_url | string | none | Last frame the clip must end on, as a public https:// URL |
keyframes | object[] | none | Up to 4 images pinned to timestamps inside the clip, each as { "image_url": "...", "timestamp": 3.0 }. timestamp is the second the image lands on, and 0 is the first frame |
voice_ids | string[] | none | 1 to 3 preset voice IDs for the spoken audio, for example ara. No repeats. Requires audio: true |
audio | boolean | true | Generate the soundtrack. false renders a silent clip |
nsfw_check | boolean | false | Check the prompt and input images for NSFW content before generating |
Unknown request fields are rejected rather than ignored, so a misspelled parameter fails the request. The Grok Imagine 1.5 Video API documentation has the full reference.
Key features of Grok Imagine 1.5 Video
Near the top of the image-to-video Arena
Ranked third on the Arena image-to-video board as of June 2026, ahead of every Veo 3.1 variant and the highest-placed xAI model. Boards move, so read it as a current snapshot.
Text, image or references
A prompt alone is a complete request. Or animate one start frame, or hand the model up to 14 reference images to lock subject, product and setting while the prompt drives camera and action.
End frame and keyframes
Pin the last frame with end_image_url, or place up to four images at exact seconds with keyframes, so a shot lands where the edit needs it.
Preset voices and an audio switch
Sound effects, ambience and lip-synced dialogue generate with the video. Choose up to three preset voices for the spoken lines, or turn audio off for a silent clip.
Native 1080p, wide ratios
Text- and image-to-video render 1080p natively, with 21:9 and 5:2 among the aspect ratios. Reference-to-video renders at up to 720p.
Up to 15 seconds in one call
Enough for a full shot without chaining. Continue any result by URL with Grok Imagine 1.0 Video Extend when a sequence needs more.
What Grok Imagine 1.5 Video is best for
Product photo to motion ad
Pass a packaging or hero shot as the start image and animate it into a short video with matching sound for paid social.
Multi-reference brand spots
Lock product, talent and backdrop across up to 14 references, then direct them with the prompt.
1080p final cuts
Deliverables that must hold up at full resolution, rendered natively rather than upscaled.
Talking-portrait video
Animate a still portrait into a scene where the subject speaks a line in a chosen preset voice, lip-synced to the motion.
Shots that must end on a frame
Use the end frame or keyframes when the edit already knows where the shot has to land.
Finishing a draft
Explore direction on 1.5 Lite Video, then re-run the winner here with references and native 1080p.
Grok Imagine 1.5 Video use cases
Start with a clean still and let the prompt say what moves and what it sounds like. A cosmetics brand can pass a packaging shot and get a slow orbit with the product locked in frame and a soft ambient bed; a studio can animate a glass render so surfaces ripple and refract. With references instead of a start frame, the same brand can lock the product, a presenter and a location separately and let the prompt choreograph them, with a preset voice reading the line.
A practical split is to draft on Grok Imagine 1.5 Lite Video and finish here. Lite is the cheapest Grok video and takes the same prompts; once a direction is right, add references, an end frame and 1080p here for the final version, then continue it by URL with Grok Imagine 1.0 Video Extend if the cut needs more seconds.
Grok Imagine 1.5 Video limitations
It runs on xAI's API, so it is safe-for-work only, with no style presets and no saved characters, props or locations; those are on the Fast build. A start image and reference images are mutually exclusive. Reference-to-video renders at up to 720p; 1080p needs a prompt or a single start image. Clips cap at 15 seconds per request. Negative prompts are ignored, so describe what you want rather than what to avoid. The task-ID based Extend and Upscale do not accept tasks from this route; use the URL-based 1.0 Video Extend instead.
Grok Imagine 1.5 Video vs Grok Imagine 1.5 Fast Video
Both run Grok Imagine 1.5. This is the official API build; Fast is the grok.com build of the same model. Both generate sound in the same pass.
| Grok Imagine 1.5 Video | Grok Imagine 1.5 Fast Video | |
|---|---|---|
| Build | Official API | grok.com |
| Text-to-video | Yes | Yes |
| Duration | 1–15s | 1–30s |
| Resolution | 480p, 720p, native 1080p (references up to 720p) | 480p, 720p, 1080p |
| Images | One start_image_url or 1–14 image_urls, plus end_image_url and keyframes | One start_image_url or 1–7 image_urls, plus characters, props, locations |
| Voices | Up to 3 preset voice_ids | One request-level voice plus character voices |
| Audio switch | audio: false | audio: false |
| NSFW and presets | No | Spicy, Fun, Normal |
| Extend | By URL with 1.0 Video Extend | By task ID with 1.5 Fast Video Extend, plus Upscale |
| Billing key | xai/grok-imagine-1.5-video | xai/grok-imagine-1.5-fast-video |
Pick this route for control and native 1080p. Pick Fast for NSFW, presets, characters, 30 second clips, or when you want to extend and upscale by task ID.
More Grok Imagine 1.5 Video API requests
The same endpoint and task flow cover every request type. Two more requests to copy:
Image-to-video at 1080p with an end frame:
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "xai/grok-imagine-1.5-video",
"input": {
"start_image_url": "https://files.unifically.com/your-start-frame.jpg",
"end_image_url": "https://files.unifically.com/your-end-frame.jpg",
"prompt": "Camera slowly zooms in while clouds drift across the sky. AUDIO: soft wind, distant birds",
"duration": 10,
"resolution": "1080p"
}
}'
Text-to-video with references and a preset voice:
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "xai/grok-imagine-1.5-video",
"input": {
"prompt": "The presenter holds up the bottle and says: this is the one. Studio lighting, slow dolly in.",
"image_urls": [
"https://files.unifically.com/your-presenter.jpg",
"https://files.unifically.com/your-product.jpg"
],
"voice_ids": ["ara"],
"aspect_ratio": "16:9",
"duration": 8,
"resolution": "720p"
}
}'
FAQs
People also ask about Grok Imagine 1.5 Video
It is the official API build of xAI's Grok Imagine 1.5 video model. Generate from a prompt alone, animate one start image, or guide the clip with up to 14 reference images, optionally pinning the last frame and adding preset voices. Clips run 1 to 15 seconds with sound generated in the same pass.
Official-build requests to xai/grok-imagine-1.5-video authenticate with a key from the API keys page in an Authorization: Bearer header. Usage is paid from a prepaid balance per second of video, and the same key checks progress at /v1/tasks/{task_id}.
An input with prompt, aspect_ratio, duration (1 to 15 seconds) and resolution up to native 1080p makes a text-to-video task; add one start_image_url or up to 14 image_urls, plus an optional end_image_url or voice_ids. The Grok Imagine 1.5 Video API returns a task_id at once, and the finished task carries the video URL.
Yes, since October 2026. A prompt alone is a complete request. Earlier the route required a start image; that limit is gone.
Either one start_image_url to animate, or 1 to 14 image_urls as references for subject, style or scene, never both in one request. An end_image_url can be added in either mode, and the API also accepts up to 4 timed keyframes.
Yes, by default. Sound effects, ambience, and lip-synced dialogue are generated with the video. Pass up to 3 preset voice_ids to choose the voices for spoken lines, or set audio to false for a silent clip.
480p, 720p, or native 1080p for text-to-video and image-to-video. Reference-to-video renders at up to 720p.
Not on this route. It runs on xAI's API, which is safe-for-work only. NSFW output, the spicy and fun presets, saved characters and 30 second clips are on Grok Imagine 1.5 Fast Video, the grok.com build of the same model.
On the Arena image-to-video board it is the highest-placed xAI model and sits ahead of every Veo 3.1 variant, third overall behind Seedance 2.0 and Gemini Omni Flash as of June 2026. Arena boards move, so treat the exact place as a current snapshot.
Yes, by URL. Pass the finished clip to Grok Imagine 1.0 Video Extend to add 2 to 10 seconds. The task-ID based Extend and Upscale routes only accept Grok Imagine 1.5 Fast Video tasks.
Same 1.5 model, two builds. This one is the official API build: native 1080p, up to 14 references, end frame, keyframes, preset voices, an audio switch and the wide 21:9 and 5:2 ratios, safe-for-work. Fast is the grok.com build: NSFW and style presets, saved characters, props and locations, clips up to 30 seconds, and the only build that Extend by task ID and Upscale accept.
Grok Imagine 1.5 Video guides and comparisons
Review pricing, limits, and tested outputs before running this model.
