Grok Imagine 2.0 Image API

What is Grok Imagine 2.0 Image?

Grok Imagine 2.0 Image is the quality-focused image model in the Grok Imagine family. Use xai/grok-imagine-2.0-image for final-quality text-to-image generation and reference-guided edits. Omit image_urls for a fresh generation, or pass 1 to 5 reference images to edit, composite, or restyle them while remaining on the same 2.0 model.

The former enable_pro input is no longer supported. Speed and quality now have separate model IDs: xai/grok-imagine-image for speed and xai/grok-imagine-2.0-image for quality.

Key features

Quality selected by model ID

Choose xai/grok-imagine-2.0-image directly. There is no Pro toggle and no hidden model switch in the request.

Text-to-image and reference editing

Generate from text alone or add 1 to 5 images for editing. References change the workflow, not the selected model.

Five production-ready aspect ratios

Generate at 1:1, 2:3, 3:2, 9:16, or 16:9 for product, social, portrait, and widescreen layouts.

Prompt enhancement when needed

Use upsample_prompt to expand a short creative brief, or leave it disabled for direct prompt control.

Best for

Final campaign assets

Use the 2.0 quality model after composition and messaging have been approved.

Reference-guided product work

Combine up to 5 references while keeping the request on the quality model.

Portraits and hero imagery

Choose 2:3, 3:2, or 16:9 for polished assets where visual fidelity matters more than iteration speed.

Two-stage workflows

Draft with xai/grok-imagine-image, then move the winning prompt and references to this model for the quality pass.

API example

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-2.0-image",
    "input": {
      "prompt": "A cinematic product portrait with detailed studio lighting",
      "aspect_ratio": "2:3"
    }
  }'

Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or provide a callback_url to receive the completed image.

FAQs

Guides and comparisons

Review pricing, limits, and tested outputs before running this model.