# Qwen-Image-2.1 Adds RGBA to Image Generation

> Alibaba's Qwen team updates its open image model with transparency support and joint text-to-image and editing capabilities.

Published by The Open Weights on Sep 14, 2026. Canonical: https://theopenweights.com/news/qwen-image-2-1-haxk

## Key facts

- Company: Qwen · Alibaba
- Model: Qwen-Image-2.1
- Version: 2.1
- Category: Text → Image
- Modalities: Text → Image, Image Editing
- License: Other (Open weights, commercial use allowed)
- Architecture: QwenImage21Pipeline
- Precision: BF16
- Weights size: 14.2 GB
- Significance: notable
- Published: 2026-09-14
- Last verified: 2026-10-23
- Hugging Face: https://huggingface.co/Qwen/Qwen-Image-2.1
- Canonical URL: https://theopenweights.com/news/qwen-image-2-1-haxk

Alibaba's Qwen team has published [Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1) on Hugging Face, an image model that handles both text-to-image generation and image editing in a single package. The headline addition this time is RGBA support, meaning the model can produce images with a genuine alpha channel rather than flat, opaque frames.

That transparency capability is more consequential than it sounds. Most open image generators output solid backgrounds, forcing designers to run a separate cutout or masking step. Native RGBA lets the model emit assets — logos, icons, product shots, UI elements — that drop cleanly onto other layers.

## Why it matters

Qwen has been steadily building out an open multimodal stack, and combining generation with editing in one model lowers the barrier for developers who want a single dependency rather than a pipeline of specialized tools. Key points from the release:

- Covers both **text-to-image** and **image editing** workflows
- Adds **RGBA / transparency** output
- Distributed openly through Hugging Face under a custom license

As always with open weights, the practical details — resolution ceilings, hardware requirements, and the exact license terms — are worth checking on the [model card](https://huggingface.co/Qwen/Qwen-Image-2.1) before building it into production. But the direction is clear: Qwen is positioning its image model as a self-contained creative tool rather than a raw research artifact.

## Get the model

- [Hugging Face](https://huggingface.co/Qwen/Qwen-Image-2.1)

## Sources

- [Qwen/Qwen-Image-2.1](https://huggingface.co/Qwen/Qwen-Image-2.1) — Hugging Face, Sep 14, 2026

---
Source: The Open Weights (https://theopenweights.com/). Aggregated and written by Claude, curated by humans. Cite as: "Qwen-Image-2.1 Adds RGBA to Image Generation", The Open Weights, Sep 14, 2026, https://theopenweights.com/news/qwen-image-2-1-haxk