Editly

How to Use Grok Imagine to Generate Images Online

Grok Imagine is xAI's image and video system, normally reached through grok.com and the Grok apps. Editly puts its text-to-image side on an ordinary web page: write a prompt, choose 1:1, 2:3 or 3:2, spend 15 credits. No subscription to xAI, no API key, nothing to install.

Start creating

How to use Grok Imagine on Editly in three steps

  1. 01

    Open the image generator

    Head to the AI image generator and select Grok Imagine from the model list. It sits next to Qwen Image 3, Nano Banana and Seedream 4.5, all drawing on one credit balance, so trying a second model later costs a click, not another signup. Sign in first if you want the result kept in your dashboard.

  2. 02

    Brief it in plain sentences

    The prompt box accepts up to 5,000 characters. xAI built Imagine to read written instructions rather than stacked keyword tags, so describe the subject, the light and the framing the way you would brief a photographer. Quote any wording you want rendered inside the picture, and say where on the canvas it belongs.

  3. 03

    Pick a shape and generate

    Three aspect ratios are wired up here: 1:1 for avatars and product tiles, 2:3 for posters and phone wallpapers, 3:2 for wide hero art. Composition follows the frame, so choose it before you write instead of cropping afterwards. Each run costs 15 credits flat and the result lands in your dashboard ready to download.

What this page actually hands you

An honest inventory, so nothing here sends you hunting for a control that does not exist.

Text to image, nothing else to configure

This page routes to Grok Imagine in text-to-image mode. Prompt in, picture out. The reference-image editing and the video modes xAI ships in its own apps and API are not connected here, and there is no hidden upload field pretending otherwise. Two fields, one button, no settings panel to decode before your first run.

Prompts up to 5,000 characters

Long, specific briefs are where this model earns its reputation. xAI says its current Imagine image model plans typography and layout the way a designer would, so spelling out the headline, the subhead and where each sits is worth the extra sentences. A paragraph of direction beats a comma salad of adjectives almost every time.

15 credits an image, no subscription

One flat price per generation, whatever the prompt length. Credit packs open at $12 for 1,000 credits, roughly 66 Grok Imagine images, with a 365-day shelf life. The same balance pays for every other model on the site, so an experiment that goes sideways has not burned a month of some plan you now have to use up.

Ratios that match real deliverables

Square for profile art and store tiles, 2:3 for posters, flyers and vertical social, 3:2 for banners and article headers. Three shapes, each a format somebody actually ships, instead of a long menu you scroll past on the way to the same square you always pick.

Grok Imagine vs the other image models on Editly

Same prompt box, same credit balance, different trade-offs. Choose by the job in front of you, not by the badge.

Grok Imagine
Qwen Image 3
Nano Banana
Credits per image15, flat15 at 1K, 25 at 2K15, flat
Inputs acceptedText prompt onlyText, or up to 3 reference imagesText, or up to 10 reference images
Aspect ratios here1:1, 2:3, 3:2Eight, including 16:9 and 21:910 presets plus auto
Prompt length5,000 characters4,500 characters5,000 characters
Where it winsYou can describe the shot in proseYou need 2K output for printYou are recombining photos you own

Three things worth knowing before your first run

Concept frames come out fastest

Prose prompt, 3:2 output, a handful of variations in one sitting. That frame is the shape storyboards, pitch decks and article headers are already built in, so the result drops into the layout without a crop. For a vertical poster, switch to 2:3 before generating rather than trimming a wide frame down and losing the composition you liked.

Rewrite the sentence, do not stack tags

When a result misses, adding six more comma-separated adjectives rarely fixes it. Restate the sentence that describes the miss instead: where the subject stands, how close the lens is, what time of day it is, what should not be in the shot. A model trained to follow instructions responds to corrections phrased as instructions.

Moderation is xAI's, not ours

Grok Imagine applies xAI's own content policy. A prompt judged out of policy comes back rejected, and that line is drawn upstream rather than by Editly, so a request behaves here much as it does in the Grok app. If you are producing work for a client, keep real people, trademarks and brand assets out of the prompt.

Grok Imagine: subscriptions, ratios and what costs what

Run your first prompt

Pick the model, write one properly descriptive sentence, and see what comes back. 15 credits a shot, no subscription behind it.

Open the image generator