How to Use Grok Imagine: App, Prompts and API (2026)

If you are working out how to use Grok Imagine, start with what it actually is: xAI’s image and video generator, built into the Grok chatbot app and website, and also available to developers as a set of models you can call from code. You type a description, it returns several images in seconds, and you can turn any of them into a short video clip.

What is Grok Imagine?

Grok Imagine is two things that share a name. The first is the Imagine feature inside the consumer Grok app, where you generate, edit and animate pictures through a chat-like interface. The second is the family of underlying models, which developers can reach through an API (an application programming interface: a way for your own code to send requests to a service and get results back).

A few terms that come up constantly:

• Text-to-image: you write a prompt (a plain-language description), and the model produces an image that matches it.
• Image-to-video: you supply a still picture as the first frame, plus a description of the motion, and the model animates it.
• Variations: Grok Imagine returns several candidate images for one prompt so you can pick the best rather than re-rolling one at a time.
• Upscaling: re-rendering a finished output at a higher resolution, for example from 720p to 1080p.

On the model side there are four pieces worth knowing:

• Grok Imagine, the image model. It returns 6 variations per text prompt, or 2 when you attach reference images.
• Grok Imagine Video 1.5, which is image-to-video only. It animates a start frame at 480p, 720p or 1080p for 1 to 15 seconds, with strong motion quality.
• Grok Imagine Video 1.0, which does both text-to-video and image-to-video, at the same resolutions, for 6 to 30 seconds.
• Grok Imagine Upscale, which raises a finished video to 720p or 1080p.

The right answer to how to use Grok Imagine depends on which door you walk through. If you want to make a few pictures for yourself, the app is the shortest path. If you want to build Grok Imagine into a product, a content pipeline or an agent, the API is the way in, and it does not require a Grok subscription.

How to Use Grok Imagine in the Grok App

The app workflow is the same on the web and on mobile, although button labels move around between updates. Here is the loop most people settle into:

1. Sign in and check your plan. Imagine is a paid feature (more on that below). If you do not see it, your account tier is the first thing to check.
2. Open Imagine. On the website it sits in the sidebar; in the mobile app it has its own tab or entry point. It is a separate space from the normal chat.
3. Write a prompt and pick a shape. Describe the subject, setting, light and style in one or two sentences. Choose portrait for phone video, landscape for YouTube or slides, square for feeds.
4. Scroll the results. Images usually come back in seconds, so the fast move is to generate, look, adjust one detail in the prompt, and generate again. Do not try to perfect the prompt before you have seen any output.
5. Turn your favourite into a video. Open the image you like and choose the option to make a video from it. Add a short motion prompt describing what moves and how the camera behaves.
6. Edit or upload your own photo. You can also start from a photo you upload, then edit it or animate it. Only do this with pictures you own, and only animate real people who have agreed to it.
7. Download what you keep. Save the files you want to your device rather than relying on the app history.

A tip that circulates widely: open a normal Grok chat, describe what you are aiming for in loose language, and ask Grok to write an Imagine prompt for you. Then paste that prompt into Imagine. It works because the chat model is good at adding the concrete details (lens, lighting, materials) that image models respond to.

Newer versions of the app also include an agent mode for Imagine, which lets you iterate by giving instructions in conversation (“make it dusk”, “move the camera lower”) instead of rewriting the whole prompt. That loop, prompt then pick then refine then animate, is the whole of how to use Grok Imagine in the app. Everything else is about writing better prompts and knowing the limits.

Is Grok Imagine free? The “no longer free” change and spicy mode

Short answer: not in the consumer app. After a scandal in early 2026 over Grok being used to produce sexualised deepfakes of real people, xAI tightened access step by step. By March 19, 2026, Imagine was fully restricted to paid SuperGrok subscribers on both the standalone Grok website and the mobile apps. That is why so many people search for “grok imagine no longer free”: guides written in 2025 describe a free tier that no longer exists.

Why can’t I use Imagine on Grok?

If Imagine is missing or greyed out, the cause is usually one of these:

• Your plan. Free accounts do not get Imagine. This is the most common reason by far.
• An old app version. Update the app; features ship to new versions first.
• Region or age checks. Some settings depend on where you are and whether your age is verified.
• Moderation. Prompts or uploads that trip content filters get refused, sometimes without much explanation. Uploaded photos of real people are checked more strictly than they used to be.
• Load. At peak times generation can queue or fail. Try again later before assuming something is wrong with your account.

What is Grok Imagine spicy mode?

“Spicy” is a setting in the consumer app that allows more suggestive content than the default. It is the reason “grok imagine spicy mode” is such a common search, and also the reason the feature drew regulatory attention. Its availability has changed several times and now depends on subscription, age verification and region. One rule does not change: creating sexual or degrading images of real people without their consent is prohibited and may be a crime where you live. If your project needs this kind of content, you need a lawyer before you need a prompt.

If you only want occasional images or clips, paying per request through an API is often cheaper than a monthly subscription. That is the practical alternative when people ask how to use Grok Imagine without signing up for SuperGrok.

Grok Imagine prompts that work for images and videos

Prompting is where most of the quality comes from, and the patterns are simple. For images, cover four things in this order:

1. Subject: what is in the frame, with one or two specific details (“a ceramic coffee mug with a chipped rim”).
2. Setting: where it is (“on a walnut desk by a rainy window”).
3. Light: the single biggest lever for mood (“soft morning light”, “hard noon sun”, “neon at night”).
4. Style or camera: “product photo, shallow depth of field”, “35mm film still”, “flat vector illustration”.

For videos, the rule is different: the start image already defines how things look, so the video prompt should describe only motion. Say what moves, how fast, and what the camera does. Re-describing the scene tends to make the model change it.

Goal	Weak prompt	Stronger prompt
Product image	a coffee mug	A ceramic coffee mug on a walnut desk by a rainy window, soft morning light, shallow depth of field, product photo
Landscape	a nice mountain	A granite peak above a pine valley at golden hour, low clouds in the valley, wide-angle, crisp detail
Illustration	cute robot	A small round robot watering a windowsill plant, flat vector illustration, pastel palette, clean outlines
Video from the mug image	a coffee mug on a desk in the rain	Steam rises slowly from the mug, rain streaks down the glass, the camera pushes in gently
Video from the landscape	make it move	Clouds drift through the valley, the light warms, slow aerial pan from left to right

Three more habits that pay off:

• Change one thing at a time. If you change subject, light and style together, you cannot tell which change helped.
• Match aspect ratio to destination early. Cropping a square image to 9:16 later throws away most of it.
• Keep videos simple. One clear motion plus one camera move beats a list of five actions. Short clips with a single idea look far more convincing.

For a deeper set of motion prompts and image-to-video examples, see our guide to Grok AI image to video. Prompting is the same whether you work in the app or through code, which makes the next section easier.

How to use Grok Imagine with the API (grok imagine api)

Learning how to use Grok Imagine from code takes about ten minutes. Through Unicorn API you get one key and one bill for the Grok Imagine models and every other image, video, audio and chat model, priced in credits. The flow for media models is always the same:

1. Create a key on the API keys page and export it as UNICORN_API_KEY.
2. Send POST /v1/jobs with a model id and an input object.
3. Poll GET /v1/jobs/{id} until status is "succeeded".
4. Read the result from job.outputs[0].url (image jobs return several outputs, one per variation).

Check the price first

Every request has an exact price you can fetch before running it, using POST /v1/models/price. Send the model and input you intend to run:

curl -X POST https://api.unicornapi.net/v1/models/price \
  -H "Authorization: Bearer $UNICORN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine-video-1-5-preview",
    "input": {
      "image": "https://example.com/start-frame.png",
      "prompt": "Steam rises slowly from the mug, the camera pushes in gently",
      "duration": 5,
      "resolution": "720p"
    }
  }'

Generate images, then animate the best one

This Python script runs the full loop: generate Grok Imagine variations in 16:9, print every URL, then animate the first one with Grok Imagine Video 1.5 for 5 seconds at 720p.

import os
import time
import requests

API = "https://api.unicornapi.net"
HEADERS = {"Authorization": f"Bearer {os.environ['UNICORN_API_KEY']}"}


def run_job(model, input_data, poll_seconds=3):
    """Create a job, then poll until it succeeds or fails."""
    r = requests.post(
        f"{API}/v1/jobs",
        headers=HEADERS,
        json={"model": model, "input": input_data},
        timeout=30,
    )
    r.raise_for_status()
    job_id = r.json()["id"]

    while True:
        r = requests.get(f"{API}/v1/jobs/{job_id}", headers=HEADERS, timeout=30)
        r.raise_for_status()
        job = r.json()
        if job["status"] == "succeeded":
            return job
        if job["status"] in ("failed", "canceled", "cancelled"):
            raise RuntimeError(f"Job {job_id} ended with status {job['status']}")
        time.sleep(poll_seconds)


# 1. Generate image variations
images = run_job("grok-imagine", {
    "prompt": (
        "A ceramic coffee mug on a walnut desk by a rainy window, "
        "soft morning light, shallow depth of field, product photo"
    ),
    "aspect_ratio": "16:9",
})

for i, out in enumerate(images["outputs"]):
    print(f"variation {i}: {out['url']}")

start_frame = images["outputs"][0]["url"]

# 2. Animate the chosen image (Video 1.5 is image-to-video only)
video = run_job("grok-imagine-video-1-5-preview", {
    "image": start_frame,
    "prompt": "Steam rises slowly from the mug, rain streaks the glass, the camera pushes in gently",
    "duration": 5,
    "resolution": "720p",
    "aspect_ratio": "16:9",
})

print("video:", video["outputs"][0]["url"])

In a real tool you would show the variations to a person (or a vision model) and let them choose, rather than always taking index 0. To steer the look toward an existing brand image, pass URLs in ref_images; the model then returns 2 variations instead of 6.

Text-to-video, longer clips and upscaling

Grok Imagine Video 1.5 needs a start image. If you want video straight from text, or a clip longer than 15 seconds, use Grok Imagine Video 1.0 ("model": "grok-imagine/image-to-video"), which accepts a prompt alone and runs from 6 to 30 seconds. To raise a finished clip’s resolution, send a job to grok-imagine-upscale with the earlier task’s id in task_id and "resolution": "1080p". The API documentation covers every input field, and you can test prompts without code in the browser playground. If you work in Cursor or Claude Code, the MCP setup for agents lets your coding assistant call these models directly.

How much Grok Imagine costs per image and per video

Unicorn API prices everything in credits. Monthly plans give credits at $0.01 each (Starter $25, Creator $50, Studio $100, Pro $200 per month), and top-ups cost $0.015 per credit and never expire. Fractional prices round up to whole credits. Full details are on the pricing page.

Request	Credits	At plan rate	At top-up rate
Grok Imagine image request (listed default)	4	$0.04	$0.06
Video 1.5, 5 s, 480p	12	$0.12	$0.18
Video 1.5, 5 s, 720p	23 (22.5 rounded up)	$0.23	$0.345
Video 1.5, 10 s, 1080p	80	$0.80	$1.20
Video 1.0, 6 s, 720p	27	$0.27	$0.405
Video 1.0, 30 s, 1080p	240	$2.40	$3.60
Upscale to 720p / 1080p	10 / 20	$0.10 / $0.20	$0.15 / $0.30

A note on the image price: the listing describes a flat price per prompt, at 4 credits, multiplied by n when you set that field. Because the multiplier interacts with the number of variations, run the price endpoint once for your exact request and you will know precisely what a batch costs.

Video cost scales linearly with seconds and steeply with resolution: 1080p is about 3.3x the cost of 480p at the same length. That shapes the sensible workflow, covered in the mistakes section below.

A worked month

Say a small shop makes 20 product clips a month. For each one it runs one Grok Imagine batch at 4 credits, then one 5-second 720p clip with Video 1.5 at 23 credits. That is 27 credits per product, or 540 credits for the month: about $5.40 at the plan rate. The Starter plan’s $25 buys 2,500 credits, which leaves plenty of room for re-rolls and a few 1080p finals.

How it compares with other image models

Model	Price	Best for
Grok Imagine	4 credits	Fast exploration: many variations from one prompt, then animate with Grok video
Nano Banana 2 Lite	4 credits per image	Cheap 1K images and edits with up to 10 reference images, wide aspect ratios including 21:9
GPT Image 1.5	3 / 4 / 22 credits (low / medium / high)	Strict prompt adherence, legible text, transparent backgrounds

If your image needs readable text, such as a poster headline, GPT Image 1.5 is the safer pick. If you need to combine many reference photos, Nano Banana 2 Lite handles up to 10. You can generate the start frame with any of them and still animate it with Grok Imagine Video, since the video models just need an image URL. The full model catalog lists the rest.

Common mistakes when using Grok Imagine

Most frustration with how to use Grok Imagine comes from a handful of repeat errors:

• Vague prompts. “A cool car” gives you generic results. Name the model of shot, the light and the setting.
• Re-describing the scene in a video prompt. The start image sets the look. Describe motion and camera only, or the model drifts away from your image.
• Rendering 1080p before the motion is right. Test at 480p, approve the movement, then render the final at 1080p or upscale the winner. A 5-second 480p test costs 12 credits against 40 for the same clip at 1080p.
• Using Video 1.5 for text-to-video. It is image-to-video only. Send it a start frame, or switch to Video 1.0 for prompt-only clips.
• Asking for too much in one clip. One action per clip. Stitch clips together in an editor if you need a sequence.
• Skipping the price check in automated code. A loop with a wrong duration or resolution can spend more than you intended. Call the price endpoint, or cap duration and resolution in your own code.
• Polling too aggressively. Checking job status every few seconds is plenty. Video 1.5 can also be busy at peak times, so build in patience and retries.
• Uploading photos of other people. Animating someone without consent is the fastest way to get blocked, and in many places it can break the law.
• Hot-linking outputs. Download the files you keep and store them yourself rather than relying on generated URLs inside your product.

Get these right and the rest of how to use Grok Imagine is iteration: short prompts that you refine one detail at a time, cheap low-resolution tests, and a single high-quality render at the end. Whether you do that in the Grok app or through the API comes down to volume. For a few pictures a month, the app is convenient if you already pay for SuperGrok. For anything repeated, automated or built into a product, an API key and per-request credits give you the same models with costs you can see before you press run.