Text-to-Image generates from a prompt only. Image-to-Image edits up to 3 uploaded images according to your prompt.
Describe the image you want Grok Imagine to generate or the edit you want applied to your uploaded image.
Sign in to spend tokens
Prompt for Grok Imagine is required
Your generated images will appear here
Grok Imagine Image generates or edits images with xAI's Grok Imagine model — text-to-image from a prompt, or image-to-image edits on up to 3 uploaded references. Prizmad ships it as a built-in tool at 2 tokens per run, with 14 aspect ratios and up to 2K resolution.
Grok Imagine Image is xAI's image model, available on Prizmad in two modes. Text-to-Image generates a new visual from your prompt alone — cinematic product shots, bold e-commerce hero images, lifestyle scenes. Image-to-Image takes up to 3 uploaded reference images and edits them according to your prompt, so you can rework an existing product photo or creative instead of starting from scratch.
The format flexibility is unusually wide: 14 aspect ratios from 2:1 panoramas down to 1:2 verticals, including phone-native ratios like 19.5:9 and 9:20 alongside the standard 1:1, 4:3, 16:9, and 9:16. Resolution runs at 1K or 2K, output comes as JPEG, PNG, or WebP, and each run can return up to 4 images. In Image-to-Image mode, an Auto aspect ratio option keeps the shape of your uploaded reference.
On Prizmad, Grok Imagine Image uses the same token balance as AI avatars, video, , captions, and music. There's no xAI account or API key to manage — you generate from the same token balance as everything else, and every image is yours to use commercially.
Dramatic product hero shots — energy drinks on crushed ice, serums floating over water ripples — built for store pages and landing pages.
Cinematic, high-contrast ad visuals in the exact ratio of the placement, from 1:1 feed ads to 9:16 Stories and 9:20 full-screen formats.
Upload a plain supplier photo in Image-to-Image mode and prompt a new scene around it — studio lighting, lifestyle context, seasonal styling.
Realistic lifestyle scenes — earbuds on a desk in warm morning light — that read as organic social content rather than obvious ads.
2:1 and 20:9 ratios cover web banners, email headers, and marketplace storefront strips without stretching or cropping.
Generates scroll-stopping ad visuals from a prompt alone — describe the product, lighting, composition, and mood, and Grok Imagine builds the scene.
Upload up to 3 reference images and describe the edit — Grok Imagine reworks the visuals according to your prompt instead of generating from zero.
From 2:1 and 20:9 panoramas to 9:16, 9:20, and 1:2 verticals — including phone-native 19.5:9 — so creatives fit the placement exactly, no cropping.
Choose 1K for fast everyday creatives or 2K when the asset needs extra detail for larger placements.
Download results as JPEG, PNG, or WebP — WebP is handy for lightweight landing pages, PNG when you need lossless quality.
Generate 1 to 4 variants in one run to compare compositions and pick the strongest creative for testing.
Pick a mode: Text-to-Image to generate from a prompt only, or Image-to-Image to edit up to 3 uploaded images according to your prompt.
Write your prompt. Describe the product, lighting, composition, and mood — for example: a high-energy drink hero image on crushed ice with dramatic neon rim light and bold e-commerce composition.
In Image-to-Image mode, upload up to 3 reference images (PNG, JPEG, or WebP) that you want Grok Imagine to edit.
Choose an aspect ratio. Text-to-Image defaults to 1:1; Image-to-Image defaults to Auto, which follows your reference. Fourteen ratios cover everything from 2:1 banners to 9:20 full-screen verticals.
Set the resolution (1K or 2K), the output format (JPEG, PNG, or WebP), and the number of images (1–4).
Click Generate. Results typically arrive in about 2 minutes and are saved to your asset library for download and reuse.
Grok Imagine Image is listed at 2 tokens per run, in both Text-to-Image and Image-to-Image modes. Tokens come from your token balance.
Grok Imagine Image is unlocked on the same token balance you use for AI avatars, video, voiceover, and music — no xAI account, no API key setup, no separate billing.
If your tokens run dry mid-campaign, buy a one-off top-up directly from the top-up page. Every image you generate stays yours with full commercial rights.
No — each run consumes tokens from your Prizmad balance, listed at 2 tokens per generation. There's no separate xAI account or API key required; the tool is available for Prizmad tokens.
Yes. Images you generate with Grok Imagine Image on Prizmad are yours to use commercially — in paid ads, on product pages, in emails, and in printed materials.
The tool is listed at 2 tokens per run. Both modes — Text-to-Image and Image-to-Image — are available at the same listed cost, and you can request 1 to 4 images per run.
Yes. Switch to Image-to-Image mode and upload up to 3 reference images (PNG, JPEG, or WebP). Describe the edit in your prompt — a new background, different lighting, a lifestyle scene around the product — and Grok Imagine applies it. The Auto aspect ratio option keeps your reference's original shape.
Fourteen ratios: 2:1, 20:9, 19.5:9, 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16, 9:19.5, 9:20, and 1:2, plus Auto in Image-to-Image mode. That includes phone-native full-screen formats (19.5:9, 9:20) that most image tools don't offer.
Ideogram 4.0 and Seedream 5.0 Pro specialize in design work with typography and precise layouts. Grok Imagine Image is a general generate-and-edit tool with the widest aspect ratio range on the platform, including phone-native formats, at 2 tokens per run. If your creative depends on rendered headline text, reach for Ideogram or Seedream; for cinematic product scenes and photo reworks in unusual formats, use Grok Imagine.
Typically around 2 minutes. Generations run asynchronously, so you can queue a run and keep working — finished images appear in your asset library.
Resolution is 1K or 2K, and output can be JPEG, PNG, or WebP. Each run returns between 1 and 4 images, depending on the count you set.
Published 2026-07-16 · Last updated 2026-07-16