Short answer
For still-image generation and editing, xAI’s developer documentation identifies grok-imagine-image-2.0. It generates images from text prompts and can edit images using reference images. xAI documents a generation API and per-image pricing, but the documentation reviewed here does not establish whether this model is designated preview or generally available—or whether it meets your production requirements. (xAI Imagine documentation)
Keep the tasks separate: Image 2.0 is for still-image generation and editing; grok-imagine-video-1.5 is a separate video model; and grok-4.7 image understanding analyzes an image supplied as input rather than generating one. (xAI Imagine documentation)
What the still-image model does
The documented model can create an image from a text prompt or edit a supplied image using a natural-language instruction. xAI documents controls including output count, aspect ratio, resolution, quality, and response format; editing supports up to five reference images. These capabilities make it a candidate to evaluate for prompted stills, variations, and iterative edits, but the feature documentation is not a comparative image-quality benchmark. (xAI Imagine documentation)
Generated images are returned as temporary URLs by default; xAI also documents base64 output. Download URL results promptly or use base64 when you need to retain the image. (xAI image-generation documentation)
Production API: endpoint, example, and price
The documented generation endpoint is POST https://api.x.ai/v1/images/generations. The example below shows the endpoint and request body only. The generation-reference excerpt available for this guide does not establish the required authentication header, so add authentication as specified by the current xAI generation documentation before using it. (xAI Images API reference)
curl -X POST https://api.x.ai/v1/images/generations \
-H "Content-Type: application/json" \
-d '{
"model": "grok-imagine-image-2.0",
"prompt": "A matte sage-green ceramic pour-over dripper on a glass carafe, on a pale oak counter; three-quarter view, soft morning light"
}'
The reference describes a response containing a data array of generated image objects. The exact field to use to retrieve an image URL is not established by the available excerpt; consult the current response schema rather than assuming a field name. The image-generation guide says URLs are temporary, so download or process URL results promptly. (xAI Images API reference; xAI image-generation documentation)
Published API prices are per image. The Image 2.0 model page lists: 1K low, $0.04; 1.5K low, $0.05; 2K low, $0.06; 1K medium, $0.06; 1.5K medium, $0.07; and 2K medium, $0.08. Image input for editing is listed at $0.01 per input image; the generated output is also charged at its applicable per-image rate. xAI says image generation has a flat per-image price regardless of prompt length. (xAI Grok Imagine Image 2.0 model page; xAI Imagine documentation)
For example, 25 generated images at 1K low quality would cost $1.00 in generation charges, before any editing-input charges. The documented price table also lists 1.5K output. (xAI Grok Imagine Image 2.0 model page)
The model page lists the service regions us-east-1 and us-west-2, and a limit of six requests per second. These are service-region and request-rate details, not confirmation of customer-country eligibility; xAI notes that API limits can vary by team tier, so check the current limits for your account. (xAI Grok Imagine Image 2.0 model page)
Release-status caveat: xAI’s developer materials document the model and API, but the model page retrieved here does not identify Image 2.0 as preview or generally available. API documentation alone does not establish an uptime commitment, service-level agreement, or production suitability. Confirm the status and account-specific terms with xAI before relying on it for a production workload.
How to decide whether it fits
Choose Image 2.0 for a pilot if you need text-to-image generation or image editing and its documented controls match your workflow. Keep the decision open if you need a benchmark-proven winner, a contractual reliability commitment, or a specific capability the documentation does not confirm.
For a practical comparison, run the same prompts through Image 2.0 and your alternatives, keeping aspect ratio, resolution, and quality settings consistent. Score outputs against your actual requirements—for example, subject accuracy, layout, legible text if relevant, edit fidelity, acceptable variation, and cost per usable image. For editing, use the same source image and requested change. This is a proposed evaluation method, not a reported head-to-head test.
Don’t confuse it with video or image analysis
xAI lists grok-imagine-video-1.5 separately from grok-imagine-image-2.0. Video generation is a different task with separate pricing; the still-image prices above do not apply to it. (xAI Imagine documentation)
Similarly, xAI’s grok-4.7 image-understanding documentation describes sending an image for analysis, not generating a new image. Earlier third-party reports use names such as “Grok Imagine 0.1” and “Grok Imagine Video 1.5”; use the current official still-image model ID above when referring to the documented generation API. (xAI models documentation)
Bottom line: For still-image generation or editing, the documented model to evaluate is grok-imagine-image-2.0, using POST /v1/images/generations for generation. The API contract and per-image prices are documented, but the retrieved model page does not establish its preview/GA designation or production suitability. Test it against your own output requirements and verify account terms before making it a production dependency.