AI Expertise
Open Models / Qwen-Image 2.1
Q

Qwen-Image 2.1

Visit website

Description

Qwen-Image 2.1 is an open model from the Qwen team (Alibaba) that both generates and edits images. According to the author, it supports transparent backgrounds (RGBA) straight from a text prompt, up to 10 reference images, local edits by circle, markup or mask, improved text rendering, and resolutions up to 2048x2048 and wider. A faster Turbo variant is published separately. The license is the Qwen Research License: it grants rights for non-commercial purposes only (research or evaluation), and commercial use requires a separate license from Alibaba. The author provides an official demo on Hugging Face.

Pros and cons

Pros

  • Generation and editing in one model
  • Transparent background from a text prompt
  • Up to 10 reference images
  • Local edits by circle, markup or mask

Cons

  • Non-commercial license: commercial use needs a separate license from Alibaba
  • Hardware requirements are not stated by the author

Key facts

License
Qwen Research License
Commercial use
Not allowed
Developer
Qwen (Alibaba)
Interface languages
English
Checked on
Oct 16, 2026

Editorial score

How strongly the AI Expertise editors recommend this open model for each of its tasks right now, on a 0–10 scale. The score combines output quality, capabilities, ease of launch, license openness and freshness.

Images

7.2 / 10
Output quality 9.0
Low High
Features 9.5
Basic Rich
Ease of launch 6.0
Hard Easy
License openness 2.0
Restricted Open
Freshness 10.0
Outdated Fresh