Qwen-Agent

About

Qwen-Agent is Alibaba's framework for building LLM applications that use tools, plan, and remember — the same code that powers Qwen Chat in production. Its building blocks are a base chat model abstraction and composable agents that combine function calling, RAG over million-token documents, a Docker-sandboxed code interpreter, and browser automation through the BrowserQwen Chrome extension. Unlike coding-specific harnesses, it treats code execution as one tool among many: the framework is equally at home building assistants, document QA pipelines, or browser agents, with prompts and tool-call templates tuned for Qwen models though it runs against any OpenAI-compatible endpoint. The Qwen team maintains it actively, and it ships its own DeepPlanning benchmark for evaluating agent planning. Developers building Qwen-powered applications use it as the foundation layer rather than as a terminal coding tool.

Other entries by QwenLM

• Qwen Code — agent