Grok Imagine Image API: xAI Image Generator

Обновлено 18 окт. 2026 г.

What is Grok Imagine Image?

The Grok image API on UnificAlly is xai/grok-imagine-1.0-image: $0.01 per image for text-to-image or reference-guided edits with up to 5 images, in five aspect ratios, with one API key. The key was renamed from xai/grok-imagine-image on 18 October 2026 to match xai/grok-imagine-2.0-image; the old key is no longer accepted, and parameters and prices did not change.

TL;DR: Grok Imagine 1.0 Image is xAI's speed-focused image model. Omit image_urls for text-to-image, pass 1 to 5 reference URLs for edit mode, pick 1:1, 2:3, 3:2, 9:16 or 16:9, and pay $0.01 per image. For quality-first output and typography-heavy work, call xai/grok-imagine-2.0-image instead; the request body is the same.

Grok Imagine Image is the speed-focused image generation route of xAI's Grok Imagine model. One endpoint covers both text-to-image and reference-guided edits: omit image_urls and the model generates from the prompt alone, or pass 1 to 5 reference URLs and the same model switches into edit mode anchored to those visuals. Output covers five aspect ratios (1:1, 2:3, 3:2, 9:16, 16:9). For quality-focused output, select the separate xai/grok-imagine-2.0-image model. The old enable_pro input is no longer supported.

Key features of Grok Imagine Image

Five features cover what you'll use in a real production loop.

Text-to-image and reference edit on one endpoint

Skip image_urls for fresh text-to-image. Pass 1 to 5 references and the same call switches into edit mode. No conditional routing in your client code, no second model id to manage.

Speed-focused iteration

Use xai/grok-imagine-1.0-image for quick drafts and rapid creative exploration. Quality is now selected with the separate Grok Imagine 2.0 Image model instead of a request toggle.

A separate quality model

Move final-quality runs to xai/grok-imagine-2.0-image. Both model IDs support text-to-image and reference editing, so adding references never changes your selected quality tier.

Up to 5 references per edit call

Edit mode takes 1 to 5 reference images to composite subjects, transfer a style, or build a scene. Anchor product geometry while rotating background or palette across runs, or carry character identity across a campaign.

Five aspect ratios that match Grok Imagine video

1:1, 2:3, 3:2, 9:16, and 16:9. Same set as the video model, so a still mocked at 9:16 maps cleanly to a vertical video without re-prompting per ratio.

What Grok Imagine 1.0 Image is best for

Single endpoint, two workflows
Text-to-image and reference edit run on the same call. The presence of image_urls flips the workflow with no conditional routing in your client.
Two-track production runs
Draft quickly with xai/grok-imagine-1.0-image, then send the winning prompt to xai/grok-imagine-2.0-image for the quality pass.
Multi-reference colourway swaps
Up to 5 references per call lets you anchor product geometry while rotating background or palette across runs. Useful for catalogue spreads.
Fast concept exploration
Use the speed model to test compositions, palettes, and reference combinations before committing to a final-quality 2.0 render.
Cross-platform social sizing
The same prompt across 1:1, 9:16, and 16:9 generates a matched set without re-prompting per ratio. Pair with the Grok Imagine video model for consistent stills and videos.
NSFW workflows on permitted accounts
A single enable_nsfw flag handles policy gating with no extra integration. Useful for moderated adult-content pipelines that need one supported provider.

Model choice

Speed and quality are separate models. enable_pro has been removed.

Speed

Use xai/grok-imagine-1.0-image for brainstorming, bulk drafts, and rapid iteration.

Quality

Use xai/grok-imagine-2.0-image for Grok Imagine 2.0 quality. It accepts the same prompt, aspect-ratio, and reference-image inputs.

Text-to-image

Omit image_urls. The model generates from the prompt alone. Good for fresh concept work where no reference exists yet.

Reference-guided edit

Pass 1 to 5 references via image_urls. The call switches to edit mode, anchoring the result to those visuals, and returns a single image. Useful for product photography, character continuity, and background swaps.

Grok Imagine 1.0 Image use cases

Run a brainstorming loop with xai/grok-imagine-1.0-image at 1:1 to pick a hero composition, then re-run the winner with xai/grok-imagine-2.0-image for the final delivery file. Build a colourway sheet by uploading a flat product photo as a reference; references activate edit mode without changing the chosen model. Anchor a campaign with up to 5 references per call so palette, geometry, and props stay consistent across a spread.

Grok Imagine 1.0 Image API examples

Call Grok Imagine Image from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/image/xai/grok-imagine-1.0-image.

curl -X POST https://api.unifically.com/v1/tasks \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "xai/grok-imagine-1.0-image",
    "input": {
      "prompt": "A robot dancing in a field of sunflowers",
      "aspect_ratio": "16:9"
    }
  }'

Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished image URLs.

Вопросы и ответы

Частые вопросы о Grok Imagine 1.0 Image

Grok Imagine Image is the image route of xAI's Grok Imagine model. One endpoint covers both text-to-image (when you omit references) and reference-guided edits (when you pass 1 to 5 reference images). Edit mode returns one image per call.

No. It was retired on 18 October 2026 in favour of xai/grok-imagine-1.0-image, which has the same parameters and price. A request to the old key is rejected with a message naming the new one, and nothing is charged.

References are optional. Skip image_urls for pure text-to-image. Pass 1 to 5 URLs and the call switches to edit mode, with the model anchoring the result to those visuals. It composites subjects, transfers styles, and builds a scene from the references, so it fits product photography, character continuity, and background swaps.

Use the separate xai/grok-imagine-2.0-image model. The old enable_pro input has been removed; this model ID remains the speed-focused option.

For typography-heavy work where quality is the priority, select xai/grok-imagine-2.0-image. Use this speed model for rapid drafts and iteration.

Set upsample_prompt to true and the API rewrites your text prompt before sampling, expanding short or vague briefs into a fuller description. Disable it when you want the model to take your prompt verbatim.

Five. 1:1, 2:3, 3:2, 9:16, and 16:9. Same five ratios as Grok Imagine video, so the same composition can be mocked as an image and animated as a video without re-prompting per ratio.

Only on accounts with the policy enabled. Pass enable_nsfw: true on the call and, if the account is permitted, the result skips the safe-for-work filter. Without the flag, results stay safe-for-work.

Edit mode (when image_urls is set as input) returns a single image_url. Text-to-image mode returns one or more images as image_urls.