What is Qwen, the AI model from Alibaba?

Qwen, also known as Tongyi Qianwen in Chinese, is a family of large language models developed by Alibaba Cloud, the cloud computing arm of the Chinese company Alibaba Group. Alongside DeepSeek, Qwen has become one of the most important sources of open-weight AI models in the world.

Alibaba has released many Qwen models openly, in a wide range of sizes, from small models that run on laptops and phones to very large models that compete with leading closed systems on many benchmarks. Many have been released under the permissive Apache 2.0 licence, which allows broad commercial use.

The Qwen family includes:

1. General language models for chat, writing and reasoning.
2. Coding models, designed for programming tasks and agentic coding.
3. Vision-language models that understand images and documents.
4. Audio and multimodal models.
5. Reasoning models that think step by step.
6. Embedding models for search and retrieval.

Why is Qwen so popular with developers?

Range of sizes. Developers can choose a model that matches their hardware and budget.
Strong performance. Qwen models often rank highly among open models.
Permissive licensing for many models.
Multilingual ability, covering many languages.
A huge ecosystem. Qwen models are among the most downloaded and most fine-tuned on Hugging Face, and many other models are built on top of them.

You can run Qwen through tools like Ollama and LM Studio, download it from Hugging Face or access it through Alibaba Cloud and other providers. Alibaba also offers its own consumer chat app.

Some considerations apply. Like other Chinese AI models, Qwen's hosted versions follow Chinese regulations, which can affect how they answer politically sensitive questions. Running the open model yourself avoids data being sent to external servers, though the model's training may still shape its answers on certain topics. Some organisations and governments have policies about using AI models developed in particular countries, so check your organisation's rules.

As always, check the licence for the specific model, since terms can vary between releases.

For many developers building AI features, especially on limited hardware, Qwen models are among the first to test, because they often offer excellent performance for their size.