grok-imagine-video-1.5-fast:reverse is the video generation endpoint in the xAI Grok Imagine series, suitable for turning text concepts, product photos, and storyboard frames into short videos. It supports text-to-video, image-to-video, and reference image guidance, with configurable aspect ratio, resolution, and duration, and delivers video links through asynchronous tasks, making it suitable for creative test shoots and asset production.
Clarify capacity, inputs and outputs, and invocation methods before selecting a model.
The durations and settings above describe this platform's invocation method and do not represent the native model's specifications across all usage environments.
Learn what grok-imagine-video-1.5-fast:reverse can bring to your work.
When no ready-made image is available, you can directly describe the subject, environment, action, and camera movement to generate a video. Text-to-video requires a prompt and is suitable for first exploring visual directions, then adjusting actions and composition iteratively; clearly stating the main intent of a single shot makes it easier to compare results than cramming an entire complex story into one prompt.
Image-to-video receives images through image_url and can be paired with prompts describing the desired motion. When you already have product photos, character visuals, or storyboard sketches, there is no need to rebuild the visual starting point from text alone; a list of reference images can also guide style and content, but reference guidance should not be understood as strictly locking every detail.
Generation tasks can be submitted asynchronously through async or callback_url, so business pages do not need to keep waiting for a connection. After saving task_id, you can query progress or receive callbacks, and obtain video_url when the status is succeeded; this approach is suitable for separating generation, review, and asset storage into clear processing steps.
Start with specific tasks to find where the model can be effective.
Input product images and specify display actions, shooting angles, and background atmosphere to generate dynamic clips for creative review. You can try different shot descriptions around the same asset, deliver multiple video candidates, and then have the team check whether the product appearance, motion effects, and visuals meet promotional needs.
Turn event themes or scene ideas into prompts, choose portrait, square, or landscape aspect ratios suitable for the publishing destination, and create short video candidates. It is recommended to first define the subject's actions and camera rhythm, then adjust the background and visual style; the delivered video assets can continue into editing, subtitles, and brand packaging workflows.
Use storyboard frames as input images, paired with descriptions of character actions and camera movement, to observe how static compositions perform once in motion. This is suitable for discussing shot directions before formal filming, or comparing different dynamic approaches for the same frame; results are used to support creative judgment rather than replace professional production with frame-by-frame control.
Choose based on task complexity, input materials, and expected results.
If you have not yet prepared a source image, or want to generate a clip longer than 15 seconds in one go, this model can be considered first: it supports starting from text alone, with a duration range of 6–30 seconds according to the guide. By comparison, grok-imagine-video:reverse has a range of 1–15 seconds. The choice should be based on input materials and clip length, rather than judging speed by the name alone.
grok-imagine-video-1.5:official requires an image_url, making it suitable for tasks that clearly use an image as the creative starting point; this model retains both text-to-video and image-to-video modes, making it convenient to use at different stages of material preparation. They are different call IDs, so when switching, you should reselect the duration and input combination rather than directly applying one version's workflow to the other.
From a small-scale task to production integration.
Define the goal, required inputs, and output requirements, using real business examples as a starting point.
Open the trial page, confirm the parameters supported by this entry point, then submit a small-scale task to review the results.
Keep the full model ID, use the request format specified in the documentation, and confirm billing rules on the Pricing page.
Understand output quality and capability scope before formal use.
Answers to common questions about using grok-imagine-video-1.5-fast:reverse.
Yes. This model supports text-to-video; simply provide a prompt describing the scene, without needing to submit image_url at the same time. The prompt can specify the subject, action, environment, and camera movement; if you already have a clear product or character image, you can instead use image-to-video to establish a visual starting point.
After submitting image_url for image-to-video, the prompt is optional, but it is recommended for describing the desired actions and camera changes. The image provides the visual material, while the text supplements the intended motion; for example, explain how the subject moves or how the camera approaches, rather than merely repeating content already present in the image.
image_url is used as the input image for image-to-video, while reference_image_urls is used to guide style or content. Reference images can help convey the creative direction, but should not be understood as mandatory copying or specified first and last frames; if the goal is to preserve the visual starting point of an image, prioritize image-to-video.
This model uses a duration range of 6–30 seconds, with a default of 6 seconds; the guide recommends trying 6 or 10 seconds first. Resolution settings offer 480p, 720p, and 1080p, with 480p as the default; you can first validate motion and composition, then choose other settings based on your use case and check the actual result in the final video.
After asynchronous submission, save the task_id and query the task through POST /grok/tasks, or use callback_url to receive completion notifications. pending means it is still processing, succeeded means success, and failed means failure; after success, read video_url in data to obtain the video, and do not rely solely on the submission response to determine whether the finished video is ready.
Model information · Updated: 2026-10-01. For call parameters and billing rules, see the API and pricing sections.
Start with a clear goal and judge from real results whether it fits your work.
This is only a basic call example. See the full docs for more parameters and advanced usage.
View Docs