Qwen Image 2.1 Pro: generate and edit images
Create images from a prompt, or edit and combine up to 10 reference images, in one model. Qwen Image 2.1 Pro spells quoted words as written, including mixed English and Chinese, and keeps faces and products stable while the scene changes. Draft in 1K, finish in 2K.

Why creators choose Qwen Image 2.1 Pro
Accurate text, stable identities, and one model for both generating and editing.
How to create with Qwen Image 2.1 Pro
Four steps from prompt to final image, for both text to image and editing.
Describe the subject, scene and style, up to 3,000 characters. Put any text that must appear in quotes and spell it exactly.
Upload 1 to 10 images, up to 10 MB each, to edit or combine them. Refer to them by position: first image, second image.
Pick 1K for quick drafts or 2K for the final, which is the default. Choose 16:9, 9:16, 1:1, 3:4 or 4:3, and generate 1 to 4 images per request.
Each image takes about a minute. To refine, say what to keep as well as what to change, and start each new edit from the original image.
Who is Qwen Image 2.1 Pro for
Qwen Image 2.1 Pro suits creators who need exact text, consistent products and multi-image edits. For dense, multi-panel layouts, try Qwen Image 3.0.
Try Qwen Image 2.1 Pro prompts
Quote any text that must appear and spell it exactly. Try this one: A minimalist poster for a jazz night. Headline in bold sans serif: "BLUE HOUR JAZZ". Smaller line underneath: "Fri 9 PM". A saxophone lit by warm amber light against a deep navy background, lots of negative space, 3:4.
Qwen Image 2.1 Pro capabilities
Two variants, Text to Image and Edit, with the same text and identity control across both.
Text to image
Generate from a prompt alone in 1K or 2K. Use it for posters, product shots and any scene built from a written brief.
Image editing
Qwen Image 2.1 Pro Edit takes 1 to 10 reference images and returns up to 2K. Edit one image or combine several into a new one.
Quoted text rendering
Words in quotes come out spelled as written, including mixed English and Chinese. Put headlines, labels and signs straight into the image.
Up to 2K resolution
1K for drafts, 2K for finals, for example 2688×1512 at 16:9 or 2048×2048 at 1:1. Two clear steps, no guessing.
5 aspect ratios
16:9 for thumbnails and video frames, 9:16 for stories and Reels, 1:1 for posts, 3:4 for portrait posters, 4:3 for classic landscape.
Multi-reference by position
Name each reference as the first image, the second image and so on, so the model knows which element comes from where.
Identity preservation
Faces and products stay stable while the scene changes, which keeps a character or product consistent across a set.
3,000-character prompts
Room for a full creative brief: subject, layout, style, exact text and what to keep unchanged in an edit.
1 to 4 images per request
Generate a small set in one go to compare compositions, then pick the one to finish in 2K.
References up to 10 MB each
Bring full-quality product photos and portraits as references, up to 10 MB per image, 10 images per edit.
See what's possible with Qwen Image 2.1 Pro
Posters, product shots, multi-reference edits and social formats, all made with Qwen Image 2.1 Pro. Use them as a starting point for your own images.

Jazz concert poster with the title Blue Hour Jazz and a saxophone, generated with Qwen Image 2.1 Pro

Amber dropper bottle with a legible DUSK Hand Oil label on a clean studio background

Person in a yellow jacket, outfit swapped using two reference images with the face and pose preserved

Vertical 9:16 social story graphic with a title and caption rendered as legible text

Storefront at dusk with a sign in English and Chinese, rendered with legible text by Qwen Image 2.1 Pro

The same DUSK hand oil bottle placed in three new scenes: marble sink, lamplit desk and sunny windowsill

Interior scene composed from many reference images: plant, camera, stool, lamp, poster and a person
Qwen Image 2.1 Pro FAQ
Qwen Image 2.1 Pro is the top tier of Alibaba's Qwen Image 2.1 family. One model handles both text-to-image generation and image editing, with accurate text rendering and strong identity preservation.
Posters, product shots and multi-reference editing. It handles a headline, a product and a background without crowding them, and keeps a face or product stable while the scene changes.
Yes. The Edit variant takes 1 to 10 reference images, up to 10 MB each, and outputs up to 2K. Refer to each image by position, such as the first image or the second image.
Yes. Put the text in quotes and spell it exactly. Quoted words come out as written, including mixed English and Chinese, so headlines and labels can go straight into the image.
1K and 2K, for example 2688×1512 at 16:9 or 2048×2048 at 1:1. Aspect ratios are 16:9, 9:16, 1:1, 3:4 and 4:3. You can generate 1 to 4 images per request.
Qwen Image 2.1 Pro is the stronger pick for posters, product shots and editing with up to 10 references. Qwen Image 3.0 handles dense, multi-panel layouts better.
About a minute per image. Use 1K for quick drafts to test a prompt, then switch to 2K for the final render.
Say what to keep as well as what to change, and start each new edit from the original image. Chained edits can drift over time, so returning to the original keeps faces stable.
Alibaba. It is part of the Qwen Image 2.1 family and is available in Artlist's AI toolkit alongside other image models.
Still have questions? We're here to help.

