Qwen-Image-2.1: Alibaba's image model adds true transparency, but a stricter license
Alibaba has released Qwen-Image-2.1, an image model for generating and editing pictures that also runs locally. It supports true transparent backgrounds, high resolutions and up to 10 reference images. The catch: the license only allows research and personal testing.
What it's about
Alibaba's AI group has released a new image model, Qwen-Image-2.1. According to heise, it came out on September 20, 2026, and you can download it for free, including via Hugging Face. Hugging Face also hosts an online demo (a Space) that you can try out. It already has 484 likes. The model creates new images from text descriptions and edits existing images, both in a single model.
What the model is supposed to do
The release notes list several features that should appeal to heavy users of image AI:
- True transparent backgrounds: Images can be generated with an alpha channel. If the model delivers on its promise, you no longer need cutting out subjects as a separate step.
- High resolutions.
- Local image editing: Change specific regions of an image without touching the rest.
- Up to 10 reference images: These are example images the result should be based on.
As examples, Alibaba showed empty rooms furnished with pieces specified through reference images, convincingly realistic group photos assembled from individual photos, plus infographics and app interfaces. These are the vendor's own showcase examples.
Comparison and open questions
Qwen says it outperforms Google's image model "Nano Banana 2" in its own tests. The source cites no independent confirmation, and heise explicitly says that tests will first have to show how good the model really is. It's also unclear how it handles German or, more generally, non-Chinese text in images. No information is available.
New license
The biggest point of criticism is the license. Earlier Qwen models were released under Apache 2.0, which allows free use, including commercial use. Qwen-Image-2.1, by contrast, is under a research license. It permits personal experimentation and research. If you want to use it in your own product or service, you need a separate commercial agreement with Alibaba. Even the text model Qwen3.8-27B, which heise calls a strong local alternative to top models from OpenAI and Anthropic, was still released under Apache 2.0.
What you need to try it
The source gives no hardware requirements. The easiest way in is the free demo on Hugging Face. For running it locally, image models of this class generally call for a graphics card with plenty of VRAM (video memory). How much exactly is unclear here and depends on model size and quantization. Check the Model Card on Hugging Face. Also plan for several gigabytes of storage.
Our take
Technically, the package sounds very appealing, especially the native transparency and multiple references. It's free to use for hobby projects and research. For companies, freelancers with paid jobs and service operators, the license is an obstacle. The performance claims come from the vendor, and independent tests still have to confirm them.
Sources
- Hugging Face Spaces huggingface.co
- heise.de heise.de
Try it yourself: Qwen-Image-2.1
Explained step by step, no prior knowledge needed – on your own computer or free in your browser.
What can I do with it?
Qwen-Image-2.1 generates images from text and edits existing images in a single model. Its standout features are true transparent backgrounds, high resolutions, targeted changes to individual image regions and up to 10 reference images. You can use it, for example, to furnish rooms with pieces from reference photos or to build group photos from individual shots. Infographics and app interfaces are also said to work. Alibaba says it beats Google's Nano Banana 2, but that hasn't been independently verified. It's unclear how well German text in images works.
Who is it worth it for?
Worth a look for hobby creatives, designers testing privately and researchers looking for a local image model with transparency. If you want to use the model in a commercial product, you'll need an agreement with Alibaba, so you should probably steer clear.
Ideas: What you can do with it
Have products generated directly with a transparent background and use them in presentations or craft projects. It saves you the manual cutout.
Upload a photo of an empty room and reference images of furniture. The model fills the room to show a furnishing idea.
Merge portraits of friends or family into one shared picture, for an invitation card, say.
Generate interface mockups for an app idea to show colleagues or friends.
Change only individual regions, such as the sky or background, without altering the rest of the photo.
What your computer needs
The source mentions no variants or quantizations. Start with the Hugging Face demo. For local use, check the Model Card and, if needed, look for a quantized community version.
- The source gives no VRAM figures. Image models of this class generally need a GPU with a lot of VRAM.
- The Hugging Face demo runs for free in your browser (a ZeroGPU Space) and needs no hardware of your own.
- High resolutions and multiple reference images take up extra memory.
- Plan for several gigabytes of storage for the weights.
How do I find out what my computer has?
- Windows: Press Ctrl + Shift + Esc → Task Manager → Performance tab → click GPU on the left. At the bottom you'll see Dedicated GPU memory (that's your VRAM). You'll find system memory under Memory.
- Mac: Apple menu at the top left → About This Mac. You'll see your chip (e.g. M2) and memory (e.g. 16 GB). On Apple chips, graphics and processor share this memory.
- Linux: Type
nvidia-smiin the terminal (shows your graphics card and its memory) andfree -hfor system memory.
Guide
Google Colab (in your browser, no installation)
Google Colab is a free service from Google. It gives you a computer with a graphics card in your browser for a few hours. All you need is a Google account. For this model: Colab Pro only – Free Colab has only 12.7 GB of RAM – not enough for this model.
- Download the notebook. Click the button and save the file.
- Open Colab. Go to colab.research.google.com and sign in with your Google account.
- Upload the notebook. In the window that appears, click Upload on the left (or choose File → Upload notebook at the top) and select the file
colab.ipynb. - Turn on the GPU. In the top menu, go to Runtime → Change runtime type → under Hardware accelerator, select T4 GPU (with Colab Pro: L4 or A100) → Save.
- Start everything. Runtime → Run all (keyboard shortcut Ctrl + F9). If you see the warning “This notebook was not authored by Google,” click Run anyway.
- Wait. The cells run one after another. A green checkmark appears to the left of each cell when it's done. Installation and download take about 15 minutes here.
- Try it out. The result appears below the last cell. To test something else, change the text in the cell and run it again with Shift + Enter.
- Clean up. When you're done: Runtime → Disconnect and delete runtime. This saves your free quota.
Good to know: The free GPU is limited (usually a few hours per day). After about 90 minutes of inactivity, Colab disconnects, and you'll have to start everything over.
Windows 10 / 11
This model runs on Python. Our script handles the setup automatically: it creates an isolated Python environment, installs the required packages (PyTorch, approx. 3 GB) and downloads the model (33 GB (about 44 min at 100 Mbit/s)).
- Update your graphics driver (NVIDIA): install the latest driver. Without an NVIDIA card, it runs on the CPU – much more slowly.
- Install Python. Download Python 3.12 from python.org and run the installer. Important: At the bottom of the first window, tick Add python.exe to PATH, then click Install Now.
- Create a folder and download the files. Create a folder, e.g.
Documents\AI-Test, and save both files into it: - Open a terminal in the folder. Open the folder in File Explorer, right-click an empty spot → Open in Terminal.
- Run it. Paste this command and press Enter:
Installation and download are running now. The first time, this can take 10–30 minutes.powershell -ExecutionPolicy Bypass -File try_windows.ps1 - Result. The folder now contains the generated image
output.png. - Try your own input: Open
demo.pyin an editor, change the sample text, and save. Then run this in the terminal:.\.venv-qwen-image-2-1\Scripts\Activate.ps1 python demo.py
macOS
Macs with an Apple chip (M1 to M5) are ideal. Older Intel Macs are very slow.
- Open Terminal. Press Cmd + Space, type Terminal, and press Enter.
- Install the developer tools (first time only; includes Python):
Click Install in the window. If you see "already installed," everything is already in place.xcode-select --install - Download the files – both end up in your Downloads folder:
- Run it:
The first time, installation and download take 10–30 minutes.cd ~/Downloads bash try_mac.sh - Result. The folder now contains the generated image
output.png.
Linux
- Check your GPU.
nvidia-smishould show your card. If it doesn't, install the driver on Ubuntu and restart:sudo ubuntu-drivers install - Install the Python basics (Ubuntu/Debian):
sudo apt update && sudo apt install -y python3-venv python3-pip git - Put the files in a folder and open a terminal there:
- Run it:
bash try_linux.sh - Result. The folder now contains the generated image
output.png.
Try it
A red bicycle helmet on a transparent background, product photo style.Furnish this empty living room with the furniture from the reference images.Create an infographic about the water cycle with clear icons.Change only the sky in this photo to a sunset and keep everything else the same.Combine these five individual portraits into a natural-looking group photo.Tips for better results
- For cutouts, explicitly use the transparency feature instead of cropping out the subject afterwards.
- When editing, state exactly which area should change and what should stay the same.
- Test reference images step by step: one first, then more.
If something goes wrong
The license is a research license: commercial use requires an agreement with Alibaba.
German or other non-Chinese text in images is untested and may contain errors.
Hardware requirements aren't specified. If you have too little VRAM, use the online demo first.
The comparison claims against Nano Banana 2 come from Alibaba itself.
“out of memory,” a crash, or your computer freezes
The model is too big for your VRAM (video memory). Pick a smaller variant from the table above and close other programs, especially games and browsers with lots of tabs.
It runs, but extremely slowly
Your CPU is probably doing the work instead of your graphics card. In Ollama, the PROCESSOR column of ollama ps shows how much runs on the GPU. If it says "100% CPU", update your graphics driver or pick a smaller variant that fits entirely into VRAM (video memory).
“ollama” or “python” is not found
Close the terminal window and open it again so it recognizes the newly installed program. For Python on Windows: reinstall it and check Add python.exe to PATH.
Windows: "Running scripts is disabled on this system"
Run the script with exactly the command above: powershell -ExecutionPolicy Bypass -File try_windows.ps1. This allows execution for this one time only.
Error message when loading the model (version, "trust_remote_code", unknown architecture)
Brand-new models often need the latest software version. With Ollama: update Ollama. With Python: pip install -U transformers in the activated environment. The Model Card usually lists the exact version.
Tokenstrom generates the scripts and notebook automatically. Please take a quick look before running them – they're plain, readable text. Mind the model's license: custom license (see Model Card).
Ask the AI assistant
Questions about this article, installation or background? Our assistant knows this page and runs on our own machine, not in someone else's cloud.
The Assistant is an AI (Gemma 4) and can make mistakes – please check important details against the source. Questions are used only to generate an answer and are deleted after one hour. Please don't enter any personal data.
Comments
What do you think? Already tried it? Write it here, no sign-up needed.
Comments are public. Links aren't allowed, and we delete abusive posts. You can delete your own comment at any time in this browser. Learn more in the Privacy Policy.