AI&Design
All archives
AUGUST 2026 — NOVEMBER 2026 · KST

Grok Imagine Image 2.0

Trace xAI Imagine Image 2.0 from Quality Mode to its developer API, Runway Dev and Higgsfield documentation. Compare typography, region edits, references and output settings.

Posts archived
Attachments examined
Capabilities with evidence

What changed, when

Announcement dates in KST · Select a change or open its example.

Result example attached Announcement or contextual mediaBlank lane · no qualifying announcement in this source

Workflow threads

Read the dated updates and sources
  1. · Imagine Image 2.0 becomes Quality Mode

    Generate photos, illustrations and text-heavy artwork with the new Quality Mode on the Grok web and mobile apps. The launch gallery shows curated images. The public release lists grok.com/imagine, iOS and Android.

  2. · Select a region to replace or remove

    The Grok editor offers a magic wand for local edits, segmentation to select an area, and transparent export after removing a background. The wand and segmentation describe the Grok interface. The public API route exposes prompt-directed editing.

  3. · Combine reference images in Grok Imagine

    Use multiple source images to composite subjects or preserve a consistent visual direction across a character, locations and props. The launch article describes up to five inputs in Imagine. The developer API was expanded from three to five later in August.

  4. · Smart resize recomposes an existing image

    Choose a new aspect ratio and let the model fill the target frame. The release shows the same image adapted to portrait, square and landscape layouts. These examples show the Grok UI's Smart resize workflow and its illustrated ratios.

  5. · Templates package common design workflows

    Open preconfigured starting points for product photography, posters, icons, game assets and recurring editing tasks. Template covers show the intended task. Individual templates may have different inputs.

  6. · The xAI API exposes the 2.0 model

    Call grok-imagine-image-2.0 through the image API and SDK. The launch includes a text-to-image request example. Current generation and editing docs describe separate request paths. The August 28 release changed some initial defaults.

  7. · Runway Dev adds Grok Imagine Image 2

    Runway exposes grok_imagine_image_2 for text-to-image, direct editing and compositing from reference images. Developers can request one to four images in 1K or 2K. Runway's release note lists up to three references and low or medium quality, defaulting to medium.

  8. · Auto becomes the default API quality

    Omitting quality now selects auto. The documented routing uses low quality for generation and medium for editing, with billing based on the actual setting. Pin low or medium to choose quality and image cost explicitly.

  9. · The API increases its reference limit to five images

    Supply up to five source images in one edit request. The previous API limit was three. This is the xAI API limit. The official example combines three inputs, and partner endpoints document their own limits.

  10. · The API adds 21:9 and 5:2 output

    Request 21:9 for a widescreen frame or 5:2 for a banner on generation and editing calls. This release applies to the direct xAI image API. The chart does not infer matching partner presets.

  11. · xAI announces a new route for legacy Quality requests

    The September notice schedules the legacy grok-imagine-image-quality slug to route to Imagine Image 2.0 with low quality on November 2. September 2 is the announcement date. The API change happens on November 2.

  12. · Legacy Quality and Pro requests now route to 2.0 Low

    The existing grok-imagine-image-quality and grok-imagine-image-pro identifiers now resolve to Imagine Image 2.0 with quality fixed to low. The request and response structure is retained. The gRPC response reports grok-imagine-image-2.0-low as the model. The base 1.0 route is separate.

Documentation and availability notes
  • Higgsfield API · · Higgsfield lists the exact xai/grok-imagine-image-2.0 identifier and documents an input image array, 1K and 2K resolution, and low or medium quality. This is a documentation observation as of the archive cutoff.
  • Higgsfield API model catalog · The catalog marks the model as Released on August 1, 2026. No matching dated Higgsfield announcement was located. The timeline records its confirmed documentation instead.
  • xAI API migration · The direct API aliases now send requests to 2.0 at low quality. This is an alias and quality setting for the 2.0 model family.
Coverage through November 20, 2026. Exact publisher timestamps are shown in KST; date-only notices retain the publisher's day.Back to capability timeline