To use Grok Imagine effectively, create or select a still image, review it, animate only an approved version when you need video, then inspect and store the finished file with its prompt and settings. This sequence prevents a common waste: spending video generations trying to repair a product label, composition, or identity problem that was already visible in the still.
You can use Imagine at grok.com/imagine and in the Grok iOS and Android apps. The API has separate documentation for models, inputs, settings, asynchronous requests, and pricing. An API capability does not prove that the same control appears in every consumer interface or account.
This guide focuses on a practical production workflow: choosing access, building a reliable still, selecting a video route, tracking API jobs, reviewing outputs, and clearing rights before publication. Prices and product entitlements below were checked on October 22, 2026 and can change.
Approve the still before animating it. Video is a poor place to repair unresolved composition, identity, product text, or brand details.
Choose consumer Imagine or the API
xAI describes Grok Imagine Image 2.0 as generally available in Quality Mode on grok.com/imagine and the iOS and Android apps. It is also available as the API model grok-imagine-image-2.0. Consumer subscriptions and metered API calls are separate access routes.
Use the consumer interface when a creator needs direct, exploratory work. Consider the API when an application needs explicit parameters, asynchronous status handling, repeatable inputs, or records owned by the team. The API is not evidence of a native Grok connection to a project-management tool.
As checked on October 22, 2026, xAI’s consumer pricing page listed Free at $0 per month, SuperGrok at $30 per month, and SuperGrok Plus at $100 per month. SuperGrok was listed with image and video generation. SuperGrok Plus was advertised with 1080p video, higher usage, and priority access. The page does not establish a fixed Free video quota or reset schedule, so check the live product and the specific account before planning work.
The API is metered separately by model and requested output settings. Before a campaign, record the access route, account or API owner, current plan, target resolution, and whether repeatable API parameters are required. Add a pricing verification date to the campaign record.
Build and approve the still image first
Image 2.0 supports text-to-image generation and image-to-image editing. xAI’s announcement documents localized Magic Wand edits, segmentation, background removal, Smart Resize, and combining up to five source images in one edit. The five-image limit applies to Image 2.0 multi-image editing. It is not the reference-image limit for Video 1.5.
Write prompts around observable details instead of general quality claims. A useful structure is subject and identifying details, action or pose, environment, framing and camera, lighting, style, required text, and exclusions.
Illustrative product prompt: A matte-black insulated bottle on a pale stone counter, centered three-quarter view, soft window light from the left, clean studio background. Preserve the bottle shape and visible label text.
Generate or select a still, then inspect product geometry, label text, logos, faces, hands, reflections, and framing. If identity or product details must remain stable, use an authorized source image for editing rather than relying only on a text description. Make one major change at a time and keep the original file so a reviewer can compare versions or retry without losing the starting point.
A proposed application record for a still can include an application-assigned asset ID, model, exact prompt, source asset IDs, output location, and status. These are team records, not fields xAI promises to return as a complete asset-management object.
Choose the video route and control the cost
Choose the route according to what must remain fixed:
| Route | Starting point | Use it when |
|---|---|---|
| Image-to-video | A particular image | An approved still should anchor the opening visual. |
| Reference-to-video | Images or other references | References should guide the scene without necessarily fixing the first frame. |
| Text-to-video | A written description | There is no required visual starting point. |
Video 1.5 supports clips from 1 to 15 seconds and generated audio, including lip-synced speech. Current documentation supports native 1080p in text-to-video and image-to-video modes. Reference-to-video is capped at 720p and supports up to 14 reference images and up to three voice references. These are API-documented capabilities, not a promise that every consumer account exposes the same controls.
As checked on October 22, 2026, the API price for Video 1.5 was $0.08 per second at 480p, $0.14 at 720p, and $0.25 at 1080p. A hypothetical 10-second clip therefore costs $0.80 at 480p, $1.40 at 720p, or $2.50 at 1080p before other costs. Draft at a lower-cost resolution, review the result, and reserve higher resolution for candidates that pass.
xAI’s June 2026 announcement said Video 1.5 Fast produced a six-second, 720p clip in about 25 seconds. That is an example for that model and setting, not a general rendering guarantee.
Run an API generation as a trackable job
For documented reference-image workflows, provide an accessible HTTPS URL, a base64 data URI, or a Files API file_id. A private local file path is not itself an API-accessible image. Validate the reference and format before submission.
Record the settings needed to reproduce or review the run: model, prompt, source asset IDs, duration, resolution, aspect ratio, audio setting, client request key, and submission time. A minimal illustrative application record might look like this:
{
"generation_run_id": "run-881",
"vendor_request_id": "request-id-from-api",
"model": "grok-imagine-video-1.5",
"prompt": "The bottle rotates slowly a quarter turn as the camera makes a gentle push-in; keep the label facing camera and preserve the soft left-side window light.",
"input_asset_ids": ["asset-203"],
"duration_seconds": 10,
"resolution": "720p",
"status": "submitted",
"output_url": null,
"created_at": "illustrative timestamp"
}This is an application design example, not an xAI response schema. Keep a generation run separate from the resulting asset. One run represents one submission attempt, while an asset represents one persisted output file. If a run produces multiple files, each output needs its own asset record linked back to the run.
A prompt is not a unique run ID. The same prompt may be submitted intentionally more than once. Give each attempt an application-issued generation_run_id and enforce uniqueness on tenant_id plus client_request_id, or on an external idempotency key, with a database constraint or transactional upsert. Do not rely on a lookup-then-create sequence or assume xAI deduplicates retries.
Useful validation rules are deterministic: reject a missing image, an inaccessible reference, a duration outside 1 to 15 seconds, an unsupported resolution for the selected mode, or a duplicate client request key before calling the API. If submission times out, reconcile the saved run and vendor request state before creating another attempt. The integration owner handles malformed requests, polling failures, timeouts, and API errors. The creative reviewer handles visual and audio defects.
Review motion, extensions, and the final file
For a first clip, prompt one clear action and one camera movement. Add sound or pacing instructions when they affect the result. Inspect the opening and closing frames, product or character identity, logos and text, lighting direction, background objects, reflections, hands and faces, audio continuity, and dialogue timing. Review the full-resolution image or complete video with sound, not only a thumbnail.
The Imagine API overview describes video extension as adding 2 to 10 seconds and producing a combined clip. Before extending, inspect the prior clip’s final frames, retain the original, and save each extension as a separate run and asset linked to its source. A reviewer decides whether the transition is acceptable for publication.
The API also documents first-frame, last-frame, and up to four mid-video keyframes for supported Video 1.5 modes. Do not assume those controls are available in every consumer interface. Use them only when the selected API mode and documentation support the intended workflow.
Check data handling and rights before publication
xAI’s consumer FAQ says users own inputs and outputs and may use outputs commercially, subject to applicable terms and requested attribution. Ownership does not grant permission to use another person’s face or voice, a third-party trademark, copyrighted reference material, or confidential client content. Treat source authorization and publication approval as separate checks.
Keep the data regimes distinct. The consumer FAQ says consumer Grok content may be used for training unless applicable controls are disabled or Private Chat is used where available. It says Private Chat content is not used for training and is deleted from xAI systems within 30 days. X-platform controls and terms differ from web or mobile Grok.
xAI’s enterprise FAQ says business and enterprise API data is not used for training without permission. The developer security FAQ says API requests and responses are retained for 30 days by default. Zero Data Retention or another contractual arrangement may apply. Confirm the terms governing the specific account before sending confidential material.
- Source images, voices, and other inputs are authorized.
- Permissions for recognizable people and voices are cleared or marked not applicable.
- Brand, logo, likeness, and other third-party rights have been reviewed.
- Confidentiality and client restrictions have been checked.
- Applicable terms and attribution requirements have been checked and dated.
- A named human approver has recorded the final decision.
Do not assume a watermark is always present or absent. Verify the actual output and current product behavior before commercial delivery.
Organize team review in a workspace
A team can use one ClickUp task per campaign asset batch as a proposed operating pattern, not as a native Grok connector or official template. Attach the generated file and keep the prompt, model, generation_run_id, source references, resolution, duration, and rights status in task fields or a linked manifest.
Use team-defined statuses such as Generating, Review, Revisions, and Approved. Make approval depend on content, rights, and brand checks rather than on generation success alone. ClickUp documents Proofing comments on supported attached images, videos, and PDFs, including video comments tied to timestamps. Check current plan limits and attachment eligibility because the documentation says Proofing is not available for cloud-storage attachments.
Use the workspace for feedback and approval, durable storage as the file system of record, and the generation application or API logs as the record of submitted runs. Teams that need help designing this operating pattern can review ClickUp setup and automation. That page does not establish a native Grok integration.
Frequently asked questions
Is Grok Imagine free?
The pricing page lists a Free plan, but it does not establish a fixed video quota or reset schedule. SuperGrok is explicitly listed with image and video generation. Check current access and account limits before relying on a particular entitlement.
Can Grok Imagine make 1080p video?
The API supports native 1080p in documented text-to-video and image-to-video modes. Reference-to-video is capped at 720p. SuperGrok Plus is advertised with 1080p consumer video, but the live interface and account entitlement determine what a consumer can select.
Does the API return a permanent file?
Do not assume a returned output location is durable. Copy approved files to storage controlled by your team when long-term retention is required, and keep the source run and asset records linked.
Can I use generated output commercially?
xAI’s consumer FAQ says users may use outputs commercially, subject to applicable terms and requested attribution. That permission does not clear third-party likeness, voice, trademark, copyright, confidentiality, or contractual rights.
What is the difference between image-to-video and reference-to-video?
Image-to-video starts from a particular image and is the appropriate route when that still should anchor the opening visual. Reference-to-video uses references to guide the scene without necessarily fixing the first frame. Confirm the selected mode’s resolution and reference limits before submission.