Developers wanting a strong reasoning model outside the US labs can try it in Qwen Chat; API pricing is unconfirmed.
Kenning's take
What it does
Qwen3-Max-Thinking is the reasoning version of Alibaba's Qwen3-Max, announced in January 2026. Alibaba reports benchmark scores close to GPT-5.2-Thinking, Claude Opus 4.5 and Gemini 3 Pro, with leading results on some tests such as HLE with tools and Arena-Hard v2. It is available in Qwen Chat and via Alibaba Cloud Model Studio as qwen3-max-2026-01-23.
- Flagship reasoning model in Qwen3 series
- Agent tools for search, memory and code
- Company-reported scores near GPT-5.2-Thinking
- Qwen Chat access
- API model ID qwen3-max-2026-01-23
Who it suits. People and developers who want a top Chinese reasoning model for hard questions and agent tasks.
Key facts
- Category
- AI models
- Part of
- Qwen
- Made by
- Alibaba (Qwen team)
- Based in
- China
- Released
- 27 January 2026
- How to get it
- In Qwen Chat; API through Alibaba Cloud Model Studio
- Works on
- Web, API
Reviews and write-ups
- GIGAZINE 2026-01-27
FAQ
When was Qwen3-Max-Thinking released?
Alibaba announced it in late January 2026; GIGAZINE reported it on 27 January 2026.
How do I use Qwen3-Max-Thinking?
In Qwen Chat, or through Alibaba Cloud Model Studio with the model ID qwen3-max-2026-01-23.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
MAI-UI
Alibaba Tongyi research models that operate smartphone apps by looking at the screen and tapping
Qwen-Image-2.0
Alibaba's 7-billion-parameter model that both creates and edits images, with strong text rendering at 2K
Qwen-Image-Layered
Open Alibaba model that splits an image into separate transparent layers so each part can be edited alone
Qwen-Image-3.0
Alibaba's hosted image model built for text-heavy layouts, taking prompts up to 4,500 tokens