Qwen-Agent is the official agent framework from Alibaba’s Qwen team, built to turn Qwen large language models (Qwen3 and later) into working applications. It packages the models’ instruction following, tool usage, planning and memory abilities into a Python framework with function calling, MCP support, a code interpreter and retrieval-augmented generation, and it ships with ready-made example applications such as a Browser Assistant and a Code Interpreter assistant. Qwen-Agent is also the backend that powers Qwen Chat. The open-source project has gathered more than 17,000 stars on GitHub and was still actively developed in 2026.
Instead of wiring raw model APIs into an agent loop yourself, Qwen-Agent gives you the standard building blocks: define tools, let the model call them through native function calling, connect external tool servers over the Model Context Protocol, and add planning and memory on top. A built-in code interpreter lets agents write and execute Python to solve problems, while the RAG components ground answers in your own documents. Example applications show the patterns end to end, including a browser assistant that can operate web pages and a Chrome extension, so you can start from a working agent and adapt it rather than beginning with an empty file.
Qwen-Agent is free and open source under the Apache-2.0 license. You only pay for the model access you choose to plug in, such as the Qwen API or your own locally hosted models.
Qwen-Agent is for developers building LLM applications on Qwen models who want a first-party framework for tool-using agents, complete with function calling, MCP connectivity and working reference apps.