CONFIRMED | xAI will retire its Image Quality model in November

xAI confirmed on September 2, 2026 that the grok-imagine-image-quality identifier will be retired from the API on November 2. Existing integrations will not suddenly fail: after that date, requests will continue to receive responses, but they will be served by grok-imagine-image-2.0 at low quality. This is a silent model migration, not a routine catalog cleanup.

The retirement completes the Image 2.0 transition

The decision completes the transition started with Image 2.0, released in August for more precise image generation and editing. xAI says the newer model covers the use cases served by Quality at a lower price and recommends changing the model field explicitly before the deadline. Independent analysis from AI Pricing Guru confirmed the notice, the redirect behavior and the price difference.

This is not a shutdown, but the redirect has real effects

The compatibility design avoids an immediate breaking change. Request and response shapes remain the same on the generation and editing endpoints, and the old slug will continue to resolve. The problem is that technical continuity can hide a change in output. The response model field will identify which model actually served the request, making logs essential for finding when the switch happened.

The detail that can be missed is quality

The redirect will use Image 2.0 at low quality. That choice reduces cost, but it does not necessarily reproduce the visual behavior of the Quality model a team tested and approved. xAI says the new model costs 0.01 dollar less per image at every resolution. That is a real saving, but cost per image is not the same as cost per approved asset. If rework increases, part of the advantage disappears.

What changes for API users

The practical recommendation is to replace grok-imagine-image-quality with grok-imagine-image-2.0 and set quality explicitly. Low is the cost-controlled path and is what the redirect will use. Medium spends more compute for finer detail. Auto leaves the choice to the service, currently using low for generation and medium for editing. Image 2.0 also supports up to five reference images and wider aspect ratios, useful for campaigns but not a reason to accept an untested change.

For agencies and small businesses, the urgency is operational, not dramatic

Teams using the model for mockups, ad variations, catalogs or product editing do not need to rewrite the whole application. They need to find calls that still point to the old slug, test representative prompts and images, record the returned model, measure latency and compare cost per approved asset. A small canary before November is more valuable than trusting compatibility language alone.

xAI is moving quality control to the developer

The editorial reading is clear: xAI is turning a retirement into a low-friction migration, while also moving quality control to the developer. The redirect avoids an outage and may reduce operating cost. Still, leaving the old slug in production means accepting xAI's default quality choice. For any recurring creative workflow, switching explicitly now is the safer decision.

Sources

Primary source: https://docs.x.ai/developers/release-notes<br />Official migration guide: https://docs.x.ai/developers/migration/imagine-image-quality-nov-2<br />Independent source: https://www.aipricing.guru/news/grok-imagine-image-quality-retirement-cost-impact/