Qwen Code is an open-source AI coding agent that lives in the terminal and also runs in editors, a desktop app, a Web UI and chat apps. It is agentic out of the box, with Auto-Memory, Auto-Skills, SubAgents, Agent Teams and MCP support. It supports several API protocols as well as local models through Ollama or vLLM, and the provider can be switched at runtime. The software is free and open source, and costs depend on the model provider chosen.
Professional developers who prefer the command line, open-source maintainers, teams that run automation in CI/CD, and engineers who want to avoid vendor lock-in are the natural users. Those who prefer an interface can use plugins for VS Code, Zed and JetBrains, or the desktop app. Developers building custom tools can use the TypeScript, Python and Java SDKs. Node.js 22 or newer is required for the npm install, and Homebrew and standalone installers are provided.
A developer can open a terminal in a project, run qwen and ask for an explanation of the repository, then continue with edits and tests. Scripts and CI jobs call it headlessly with qwen -p to automate reviews or batch work. Teams connect it to Telegram, DingTalk, WeChat or Feishu and start a channel so colleagues can use it from chat. Others run a daemon that serves clients over HTTP and SSE, or point it at a local model for work that must stay on their own hardware.
