xAI's Imagine Image 2.0 adds precise editing to Grok, but no API yet
xAI's new image model is live in Grok's consumer apps with region editing, five-image references, and a #2 Arena claim, but developers still have no API.
What shipped
On August 7, xAI released Imagine Image 2.0 and made it generally available as the new Quality Mode in grok.com/imagine, plus the Grok iOS and Android apps. It is positioned as an image-generation model built around precise editing, not just prompt-to-image output. 1
The workflow changes after the first generation:
- A magic wand edits the region you point at while leaving the rest untouched. 1
- Segmentation lets you select a precise area, while background removal exports a subject on transparency. 1
- Multi-reference editing accepts up to five input images in one generation, reducing the need to composite them manually. 1
- Smart resize recomposes one image across nine ratios, including portrait, square, and 16:9 formats. 1
- Templates cover recurring jobs such as product color changes, e-commerce photos, headshots, icons, character sprites, and UI kits. 1
The benchmark signal
xAI says Image 2.0 ranks second worldwide for both text-to-image generation and image editing on the Arena leaderboards, based on standings dated August 7. The cited pages are Arena's text-to-image leaderboard and image-edit leaderboard; xAI says its entries appear under the name SpaceXAI. 1
That is a useful quality signal, but it is not a clean model-only benchmark. Independent coverage from Unite.AI notes that the announcement supplies no evaluation beyond the Arena standings it cites, and identifies OpenAI's gpt-image-2 as the model above Image 2.0 in both lists. 2 xAI also gives no numeric comparison with an earlier Imagine Image version, so the practical improvement is currently easier to test in the product than to measure from the release note.
What developers can do now
The access boundary is simple: consumer users can try Quality Mode now, but API access is coming soon, with no date given. 1 The announcement also does not provide API pricing or a technical model specification. That makes Image 2.0 immediately relevant for designers and creators testing iterative edits, reference-heavy compositions, or asset variations. Developers planning an automated production pipeline still have to wait.
For hands-on evaluation, start with edits that are expensive to redo manually: preserve a product while changing its color, remove or replace a background, and keep a character consistent across several references. The Arena result justifies testing; the missing API and independent benchmark evidence are the reasons to keep the verdict provisional.
References
- 1
- 2
AI Model & Product Launch Alerts
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.
Related content
- Sign in to comment.