All insights

Image generation

What Model Does Grok Imagine Use? Aurora and API IDs

xAI announced Aurora as the code name for its image-generation model. Its documentation also lists separate image and video API model IDs. Here’s what those names establish—and what they don’t.

2 min readTalk to our API team
Illustration for What model does grok imagine use

Short answer

Aurora was the code name xAI announced for its image-generation model. For API use, xAI documentation lists grok-imagine-image-2.0 for image generation and grok-imagine-video-1.5 for video generation. (Aurora announcement; Imagine API overview)

Those names have different scopes: Aurora is the historical code name in xAI’s image-generation announcement; the two grok-imagine-… strings are documented API model IDs. xAI’s image-generation tool documentation also names grok-imagine-image-2.0 for the conversational image tool. These references don’t establish that every current Grok app experience uses the same implementation, or that image and video generation share one model. (Image-generation tool docs)

What to use

  • Generate or edit images through the direct API: grok-imagine-image-2.0
  • Generate video through the API: grok-imagine-video-1.5
  • Ask Grok to make an image in a conversation: xAI’s tool documentation identifies grok-imagine-image-2.0 for its conversational image-generation tool.

These are the identifiers listed in xAI’s documentation; check the Imagine overview for the model IDs available when you build an integration.

Generate an image with xAI’s API

This is a direct xAI API request. Set an xAI API key in the XAI_API_KEY environment variable, then submit the request:

curl -X POST https://api.x.ai/v1/images/generations \
  -H "Authorization: Bearer $XAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-imagine-image-2.0",
    "prompt": "A small red fox curled up on a mossy log in a quiet forest, soft morning light"
  }'

The documented request uses bearer-token authorization and a JSON body with model and prompt. The API returns generated images in a data array; for the default URL response, read the image URL from data[0].url. xAI says generated URLs are temporary, so download or process the image promptly. The n parameter is optional on the REST API and defaults to 1. (Images API reference; image-generation guide)

The cURL example uses the default URL response. xAI’s guide also describes base64 output, but this example does not configure it.

Is xai-grok-imagine-image the same model?

Not necessarily. xai-grok-imagine-image is the identifier in Layer’s inference endpoint. It is Layer’s API identifier—not xAI’s documented model ID—and does not establish which implementation the Grok app uses. For a direct xAI API request, use the model ID listed in xAI’s own documentation.

Bottom line: Aurora is the historical code name xAI announced for its image-generation model. xAI’s documentation lists grok-imagine-image-2.0 and grok-imagine-video-1.5 as separate API model IDs, and identifies the image ID for its conversational image-generation tool. Those documented names do not establish that every current Grok app context uses the same implementation or that image and video share one model.

Sources

Grok Image Generation Release — xAI

Imagine Overview — xAI Docs

Image Generation Tool — xAI Docs

Images API Reference — xAI Docs

Image Generation — xAI Docs

Grok Imagine Image — Layer Docs

Contact us