xAI launches Imagine Image 2.0 for creating and editing images

xAI now offers Imagine Image 2.0, a model for creating and editing images that aims to solve one of the most common problems with these tools: getting usable results without starting over several times.

It is generally available in Quality Mode at grok.com/imagine, as well as in the Grok apps for iOS and Android. You can also integrate it into other applications through the API under the name grok-imagine-image-2.0.

More control over the result

Image 2.0 is designed to follow detailed instructions and preserve important elements across different generations and edits. In practice, that means you can ask for a poster with several blocks of text, a product image with specific proportions or an illustration with a consistent style without every attempt changing completely.

Typography and element placement receive special attention. The model tries to organize text and composition as a designer would, which matters when an image includes headlines, labels, menus or grouped information.

Editing is also central to the model. Its tools include:

• Magic Wand: modifies the area you select while leaving the rest untouched.
• Segmentation: selects specific areas for more precise editing.
• Background removal: extracts an object or person with a transparent background.
• Editing with multiple references: combines up to five input images in a single generation.

This can be useful, for example, for changing the color of a piece of clothing without altering the person, preparing a product photo for an online store or combining several references without having to assemble them manually in another program.

One image for any format

You can choose the canvas proportions, and the model completes the framing. That way, the same idea can be adapted to a vertical post, a horizontal thumbnail or a square format without awkward cropping.

xAI says the model was trained to preserve fidelity in photography, design and illustration. According to the Arena Image Edit and Text-to-Image rankings consulted on August 7, 2026, Image 2.0 ranks second worldwide for text-to-image generation and image editing. In those tables, xAI's models appear under the name SpaceXAI.

Templates and consistent characters

The company is also adding templates for common tasks such as photo editing, product photos, portraits, icons and video game assets. Each template prepares the workflow and lets you provide the necessary images or instructions.

Another feature aims to maintain a visual identity across multiple images. You can generate a character, its settings and the objects it uses separately while preserving the same style, building a small visual universe from several prompts.

For you, the main change is practical: fewer manual adjustments and more control over what gets modified. Its real usefulness will depend on how well it preserves details in longer projects, especially when they involve text, multiple references or many consecutive edits. That is where we will see whether its ranking position translates into a reliable tool for daily work.