liveBook · Manning

Qwen

As explained in Build Applications with Local AI Models on a Mac

Qwen is a family of language models developed by Alibaba Cloud.

Qwen is an Alibaba Cloud family of language models to consider when comparing models for language coverage, coding, or mathematical reasoning. Its multilingual strength makes it a strong candidate for work involving Asian languages. section 12.1.3

Key releases and features include:

  • Qwen 1 launched in 2023; Qwen 2.5 followed in late 2024 with improved multilingual support. section 12.1.3
  • Qwen 3, released in April 2025, includes dense and mixture-of-experts (MoE) models, hybrid reasoning, and support for 119 languages. In MoE model names, A3B and A22B indicate active parameters per token. section 12.1.3
  • Qwen 3.5, dated February 2026, is listed as a 397B-total MoE model with 17B active parameters and a 262K context. section 12.1.3

How Build Applications with Local AI Models on a Mac covers it

How deep it goes
Lists release features and model specifications without explaining implementation.
Examples
Names Chinese, English, Japanese, and Korean among its languages and compares releases from Qwen 1 to Qwen 3.5.
What you can do after
Evaluate Qwen’s fit for multilingual, coding, or mathematical-reasoning needs when choosing a model.
How it connects
Connects Qwen with multilingual support, mixture-of-experts models, coding, and mathematical reasoning.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build Applications with Local AI Models on a Mac. The section links point to the parts of the book it draws on.