qwenlm/qwen-agent is a Python agent framework for building LLM applications on top of Qwen 3.0 or newer models. The latest release listed in the repository is v0.0.26. The repository offers features such as function calling, MCP, a code interpreter and RAG, and what exactly changed in v0.0.26 is not stated in the provided facts. The basis is the GitHub repository description and the README summary.
The framework builds applications that carry out tasks using Qwen's instruction following, tool use, and planning and memory capabilities. It provides basic components such as models and tools, along with agent components built on top of them. It also includes Browser Assistant, Code Interpreter and Custom Assistant examples. It works as a backend for Qwen Chat and supports a Gradio-based GUI as well as DashScope or a custom model service configuration. It is licensed under Apache-2.0 and implemented in Python.
However, the confirmed description states only that it is based on Qwen 3.0 or newer, and it does not say whether other model families are supported. The provided facts also contain no change log for v0.0.26 and no performance benchmarks. So rather than judging how much this version improved or how it behaves on other models, it is more accurate to view it as a framework that provides the components and examples for Qwen-based applications.