Qwen image family

Use the current Qwen image generation and editing policies.

Vizify lists Qwen Image 3.0 and 3.0 Pro for new images and Qwen Image 2.1 for edits and multi-reference work, each at 1K or 2K. The original Qwen Image and Qwen2 Image Edit still run, but the newer policies succeed them.

Qwen Image 3.0 and 3.0 Pro for generation Qwen Image 2.1 for edits, up to 10 references 1K and 2K entries on the current policies Qwen Image and Qwen2 Image Edit kept as superseded
Verified catalog facts

What Vizify exposes for this model family.

These facts are generated from the current public catalog, not copied from provider marketing.

Variants5

Qwen Image 3.0 · Qwen Image 3.0 Pro · Qwen Image 2.1 · Qwen Image · Qwen2 Image Edit

Inputs2

Text · Image

Outputs1

Image

AccessTask-based

Image generation

StatusAvailable

Reviewed 2026-09-26

Best for

Where this family fits.

Generating a new image from a prompt with up to three references on Qwen Image 3.0

Editing an existing image with explicit preservation requirements on Qwen Image 2.1

Requesting a transparent background from Qwen Image 2.1

Choosing between 1K and 2K output before a request runs

Current limits

What not to assume.

Qwen Image 3.0 and 3.0 Pro list no Auto ratio, so the frame must be chosen

Semantic edits do not guarantee pixel-identical untouched regions

The public Qwen tools do not collect a mask, even though Qwen Image 2.1 accepts one

The catalog does not support a benchmark-based ranking claim

Pick the operation first

Use Qwen Image 3.0, or 3.0 Pro, when the desired artifact is new. Use Qwen Image 2.1 when an existing image is the source of truth or when a request needs more than three references. Keeping these routes separate prevents an edit request from being treated as unconstrained generation.

Current and superseded policies

Qwen Image 3.0 and 3.0 Pro list 1K and 2K with up to three references and no Auto ratio. Qwen Image 2.1 lists 1K and 2K, up to ten references, an Auto ratio, and an opaque or transparent background. The original Qwen Image and Qwen2 Image Edit keep their 1MP entry and one image input; they still run, but the current catalog names Qwen Image 3.0 and Qwen Image 2.1 as their successors.

Evidence boundary

Availability and parameters were checked against the Vizify generation catalog on 2026-09-26. External benchmark and search-demand conclusions require separate evidence.

FAQ

Questions about Qwen Image in Vizify

Which Qwen image policies are available in Vizify?

The current catalog lists Qwen Image 3.0, Qwen Image 3.0 Pro, and Qwen Image 2.1. The original Qwen Image and Qwen2 Image Edit are still listed but superseded.

Which Qwen model does each tool use?

The Qwen AI Image Generation tool uses Qwen Image 3.0. The Qwen Image Edit tool uses Qwen Image 2.1 and requires an existing source image.

Which output entries are listed?

Qwen Image 3.0, 3.0 Pro, and 2.1 list 1K and 2K, with 1K as the default. The superseded Qwen Image and Qwen2 Image Edit list 1MP.

How many image inputs can I use?

Qwen Image 3.0 and 3.0 Pro accept up to three reference images, Qwen Image 2.1 up to ten, and the two superseded policies one each.

Which aspect ratios does Qwen Image 3.0 list?

Qwen Image 3.0 and 3.0 Pro list 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, and 21:9. Qwen Image 2.1 lists the same set plus Auto.

Can Qwen Image return a transparent background?

Qwen Image 2.1 accepts an opaque or transparent background request. Qwen Image 3.0 and 3.0 Pro do not list a background control.

Does Qwen Image 2.1 preserve untouched pixels exactly?

No. Semantic editing may reconstruct nearby pixels, so protected geometry, identity, text, and edges should be stated and reviewed.

Why are Qwen Image and Qwen2 Image Edit still listed?

Both still run in Vizify, so their reviewed facts stay on this page. For new work, Qwen Image 3.0 succeeds Qwen Image and Qwen Image 2.1 succeeds Qwen2 Image Edit.