Alibaba Launches AI Model For Advanced Programming Qwen3-Coder

Qwen3-Coder (photo: Alibaba)

JAKARTA - Alibaba officially launched Qwen3-Coder, the company's most advanced AI-programming model, designed to support the development of high-performance software.

Qwen3-Coder is claimed to be superior in a variety of AI-agentic-based programming tasks ranging from producing new codes, managing complex programming workflows, to thoroughly dgging across code bases.

This open-source Qwen3-Coder-480B-A35B-Instruct model has a total of 480 billion parameters, with 35 billion parameters enabled per token, thus providing high efficiency without compromising performance.

In addition, Alibaba also released Qwen Code as an open-source project, a command-line interface (CLI) tool that allows developers to delegate software engineering tasks to AI using natural language.

Qwen Code maximizes Qwen3-Coder potential in the real-world analytic programming scenario. This model also supports integration with the Claude Code interface, making it easier for developers to execute various programming tasks.

Trained using large datasets that include general code and text, Qwen3-Coders are specially designed to support reliable agentic programming. The model natively supports a text window of up to 256,000 tokens, and can be expanded to 1 million tokens, allowing huge database processing in one session.

With its availability as open-source, strong agency capabilities, as well as seamless compatibility with various popular developer tools and interfaces, Qwen3-Coder comes as a valuable tool for global developers in encouraging innovation in software development.

The Qwen3-Coder-480B-A35B-Instruct model is now available on Hugging Face and GitHub. Developers can also access this model via Qwen Chat or using a cost-effective API available on the Studio Model, Alibaba's generative AI development platform.