Qwen
Alibaba's AI family: the free Qwen Studio chat app and open-weight models you can download and run yourself. A free mini-course from basics to power moves.
What it nails
- Free chat app with search and deep thinking
- Open models you can run yourself
- Image creation and editing built in
Watch out for
- Free usage limits are not published
Qwen is Alibaba's family of AI models, and Qwen Studio (renamed from Qwen Chat in April 2026) is its free chat app at chat.qwen.ai and on mobile. It gives "unified access to Qwen's series of open-source and proprietary models" with search, deep thinking, image and video understanding, image generation and editing, documents, voice and Artifacts. The flagship is Qwen3.8-Max (3 August 2026); smaller Qwen3.8 models are free to download.
Start here: the basics
- Open chat.qwen.ai or the Qwen Studio app and sign in.
- Pick a model at the top. A link like chat.qwen.ai/?models=qwen3.8-max opens the flagship directly.
- Turn on Deep Thinking for hard problems and Search for current facts.
- Be clear. Alibaba: "The clearer, and more specific your task description (prompt) is, the more likely the LLM's performance will meet your expectations."
Alibaba's prompt framework
Alibaba's official guide builds a prompt from Context, Objective, Style, Tone, Audience and Response, and recommends separators like ###, === or >>> between parts.
### Context I sell handmade leather wallets online. Most buyers are men aged 25-40 who value quality over price. ### Objective Write the product description for our new slim card wallet. ### Style and tone Confident, warm, no hype words. ### Audience Busy professionals reading on their phone. ### Response Three short paragraphs, then 4 bullet points of specs.
Example by Alibaba Cloud · source ↗ · A teaching prompt built on the official framework (Context, Objective, Style, Tone, Audience, Response) and its tip to use separators like ###.
Plan a 7-day Japan itinerary for two people on a mid-range budget, balancing Tokyo, Kyoto and one smaller town. /think
Example by Qwen team · source ↗ · Qwen3 models accept "/think" and "/no_think" in your message to switch reasoning on or off.
Which model to pick
| Model | What it is | Where |
|---|---|---|
| Qwen3.8-Max | Flagship; 2.4T parameters, 95B active; context up to about 1M tokens | Qwen Studio, API (open weights since 12 August 2026) |
| Qwen3.8-27B | Small open model, Apache 2.0 | Download and run yourself |
| Qwen3.8-Flash-Next | "An early preview of the Qwen4 architecture" | Open weights (26 August 2026) |
Qwen Studio has no published paid plan. On the API, Qwen3.8-Max costs $2 per million input tokens and $6 per million output tokens, with a free quota of 1 million tokens per model for 90 days (Singapore region).
Power moves
1. Build tools and pages (Artifacts)
Describe the tool precisely and Qwen writes a working page you can use.
Build an HTML page which has an input box for accepting the URL to an image and a textarea for accepting the above style of JSON. It appends the image to the page, measures its width and height, then treats the coords in the bbox_2d as scaled from 0-1000 and scales them against the actual width and height, then it renders labelled boxes over the image.
Example by Simon Willison's Weblog · source ↗ · Made with the open Qwen3.8-27B.
2. Create and edit pictures in the chat
"Visit Qwen Chat and choose 'Image Generation'," says the Qwen team. The Qwen team shows it spelling every sign and book title correctly, and it can edit your own photos.
Bookstore window display. A sign displays "New Arrivals This Week". Below, a shelf tag with the text "Best-Selling Novels Here". To the side, a colorful poster advertises "Author Meet And Greet on Saturday" with a central portrait of the author. There are four books on the bookshelf, namely "The light between worlds" "When stars are scattered" "The slient patient" "The night circus"
Example by Qwen team · source ↗ · Made with Qwen-Image, available under "Image Generation" in the chat app.
Give the pelicans rainbow colored plumage
Example by Simon Willison's Weblog · source ↗ · This is an edit: upload your photo first. Made with Qwen-Image-Edit on chat.qwen.ai.
3. Run Qwen on your own computer
Many Qwen models are open under Apache 2.0. Small ones run on a laptop, free and offline.
Generate an SVG of a pelican riding a bicycle
Example by Simon Willison's Weblog · source ↗ · A famous informal test. Made with the open Qwen3.6-27B running locally.
Generate an SVG of a flamingo riding a unicycle
Example by Simon Willison's Weblog · source ↗ · Made with the open Qwen3.6-35B-A3B running locally.
4. Qwen Code: an AI coder in your terminal
"The open-source AI coding agent that lives in your terminal." Install it with npm install -g @qwen-code/qwen-code@latest; it works with Qwen and other models.
Qwen's thinking models can "overthink": one 27B test took 21 minutes of reasoning for a single drawing. The open Qwen3.8-Max weights are text-only and always think. Free usage limits in Qwen Studio are not published, and the best Plus and Max models run only on Alibaba's own platforms.
FAQ
Is Qwen free?
Yes. Qwen Studio has no published paid plan; many Qwen models are also free to download.
Is Qwen Studio the same as Qwen Chat?
Yes. Qwen Chat was renamed Qwen Studio in April 2026; it is still at chat.qwen.ai.
Can Qwen make images?
Yes. Choose Image Generation in the chat to create pictures, or upload a photo to edit it. Our Qwen Image guide covers the picture model.
Can I run Qwen offline?
Yes. Open models like Qwen3.8-27B run on your own hardware; bigger ones need powerful GPUs.
How do I make Qwen think harder?
Turn on Deep Thinking in the app. With Qwen3 models you can also add /think or /no_think to your message.
Sources: Qwen Studio · Qwen3.8 on GitHub · Qwen3.8-Max model card · Model Studio pricing · Prompt engineering guide