MAI-Image-2.6 Review: Reference-Led Fashion Test
MAI-Image-2.6 and MAI-Image-2.6-Flash both completed Astria's matched three-reference fashion test. The flagship returned the more controlled diagonal composition and kept the dress construction and bag recognizable. Flash produced a useful campaign frame at the same native output size, with slightly simpler accessory detail.
This replaces the earlier non-comparable qualifier. The public evidence below is reference-driven from end to end.
MAI-Image-2.6 at a glance
| Variant | What this one-output test showed | Output control |
|---|---|---|
| MAI-Image-2.6 | Strong diagonal pose, recognizable Sloane identity, intact pale-yellow dress language, and a clearly visible circular-detail bag | No shared resolution selector; 1365×768 returned |
| MAI-Image-2.6-Flash | Clean commercial framing and recognizable references, with simpler bag and surface detail | No shared resolution selector; 1365×768 returned |
One output per model is a production qualification, not a universal ranking. The selection rule was fixed in advance: publish the only returned output if it passes the visual and reference-fidelity gates.
The complete input reference set
The inputs came from Maison Urban prompt 46557199 and were copied into Astria workspace 896 (Articles). Every model below received the same Sloane identity tune, yellow sleeveless belted dress tune, and circular-detail gold bag tune.
The dress and bag masters are above 2K on the long edge. The supplied Sloane portrait is only 679×722, below our preferred 1600px source gate, and was not upscaled. Identity conclusions are therefore directional.
Prompt
Dress hero shot. An off-kilter, flash-lit photograph of the referenced woman in a dynamic, almost off-balance leaning pose against a dark wooden bar counter or pillar at night. Her body creates a diagonal line. She wears the reference dress and holds the reference bag. The strong flash casts a sharp shadow. Blurred shelves with bottles and bar lights define the out-of-focus background. Cool, edgy vibe.
The live requests use the exact Articles-workspace reference tokens. They are omitted above only to keep the semantic brief readable.
Flagship vs Flash
Both outputs pass the article-size visual gate. Neither is a literal SKU reconstruction: pleat placement, belt wrap, and the number and spacing of individual bag discs vary from the references. That is acceptable for campaign ideation only after a product review; it is not automatic PDP approval.
What changed from the earlier review
The first MAI article mixed a text-only fashion frame with a direct edit of a weak synthetic source. That made the page visually polished but unsuitable as evidence for a reference workflow. Those examples and the dependent claims are no longer public in this article or its cover.
The replacement test answers a narrower and more useful question: can both MAI variants accept the same real reference set used by the other models and produce a coherent one-shot campaign frame? In this integration test, yes.
Production guidance
- Start with the flagship when garment construction and accessory separation are the approval risk.
- Use Flash to qualify composition quickly, then check whether simplified detail is acceptable.
- Keep identity, garment, and accessory references fixed when comparing variants.
- Inspect buttons, belt geometry, seams, chain, and individual bag discs at full size.
- Use the current Astria pricing page rather than a benchmark-era rate.
Bottom line
MAI-Image-2.6 is now demonstrably compatible with this three-reference Astria workflow. The flagship is the stronger result in this single matched cell; Flash remains useful when its lower-detail interpretation is sufficient. Read the full reference-matched comparison before choosing a default.