Qwen on Berges AI

The open-weights flagship from Alibaba's Qwen team, a very large reasoning model for code and long, multi-step work. Pick it from the sidebar and it answers every turn.

Try Qwen

Qwen is the family of large language models from Alibaba's Qwen team. It spans a wide range of sizes, from small models that run on a laptop to the Max line at the top. Qwen 3.8 Max is the first Max model whose weights Alibaba has published.

Berges AI serves Qwen 3.8 Max in its deep-thinking layer. It is a very large model, so it goes deeper on hard questions and costs more units per reply. Pick Qwen from the sidebar and it answers every turn. Your choice is remembered.

The weights are published under Alibaba's own Qwen license. It reads like MIT for almost everyone; only very large products and AI services must meet extra conditions.

At a glance

Qwen specs.

Maker
Alibaba (China)
License
Qwen license, MIT-style (open weights)
Architecture
Mixture-of-Experts
Current version
Qwen 3.8 Max
Best known for
Code, long multi-step tasks, and multilingual breadth
Strengths

What Qwen is good at.

Scale

Qwen 3.8 Max is one of the largest open-weights models published. That size shows up as breadth: it knows a lot, across a lot of fields, and holds up on questions where smaller models guess.

Code and long tasks

Alibaba built the 3.8 generation for coding and for tasks that run through many steps. Planning a change, reading unfamiliar code, and keeping a long line of work on track are where it is strongest.

Many languages

The Qwen line has long been one of the strongest open-weights families outside English, particularly across Asian languages.

Reach for it when

Qwen is the right pick.

  • → Code: reading, planning, and debugging
  • → Long problems with many dependent steps
  • → Work in languages other than English
Questions

Things people ask about Qwen.

Can I use Qwen without Berges AI?

Yes. Alibaba publishes the weights openly. You can run them yourself if you have the hardware, or use any open-weights host. Berges AI is one place to chat with Qwen, with privacy as a default and encryption at rest.

When should I pick Qwen?

When the question is hard: code, a long problem with many steps, or a case where a quick answer feels shallow. Qwen is a very large model, so each reply costs more of your plan's units than a small one.

Is Qwen really open source?

Open weights, under Alibaba's own Qwen license. It lets you use, modify, host, and sell products built on the model, like MIT. The extra conditions apply only to very large businesses: products with over 100 million monthly users or $20 million in monthly revenue must show the model name, and AI service companies above $50 million a year need a separate license.