Alibaba Group has launched Qwen3-Coder, an open-source AI model designed for software development.

According to the Chinese technology giant, the model is specifically tailored for tasks such as code generation and managing intricate coding workflows.

Alibaba highlighted that Qwen3-Coder excels in “agentic AI coding tasks”, which involve automated processes where AI systems address programming challenges autonomously.

Performance data from Alibaba indicates that Qwen3-Coder has surpassed domestic rivals such as DeepSeek and Moonshot AI’s K2 in core coding competencies.

Furthermore, Qwen3-Coder is reportedly on par with leading US models, including Anthropic’s Claude and OpenAI’s GPT-4, in specific areas.

The model is available in various sizes, with the most powerful version being Qwen3-Coder-480B-A35B-Instruct. This variant features 480 billion parameters and 35 billion active parameters, setting new benchmarks among open models for Agentic Coding, Agentic Browser-Use, and Agentic Tool-Use.

Alibaba claimed that the model provides long-context understanding and generation with a context length of 256,000 tokens, extendable up to one million tokens using Yarn. The model supports 358 coding languages and is optimised for repository-scale comprehension.

In parallel, Alibaba subsidiary Alibaba Cloud is expanding its collaboration with HP to develop AI PCs in China. This partnership involves integrating Alibaba’s open-source large language model, Qwen, with three billion parameters, into HP’s smart assistant “Xiaowei Hui”.

This integration aims to enhance the AI PC experience by offering functionalities such as smart document drafting and intelligent meeting note summarisation.

Qwen has been deployed across industries such as automotive, smartphones, robotics, and home appliances.

It powers advanced AI features for various sectors, including robotics, automakers and home appliances.