Alibaba introduced Qwen3-Coder, a groundbreaking open-source AI model tailored for software development and autonomous programming. The advanced version features a 480-billion-parameter Mixture-of-Experts architecture with 35 billion active parameters and a context window expandable up to one million tokens. It purportedly sets new benchmarks in performance for agentic coding and has comparable capabilities to Anthropic's Claude Sonnet 4. Alongside, a command-line tool named Qwen Code has been released, optimized for agentic coding and compatible with existing developer tools, reflecting a growing trend in AI innovation within China.
Alibaba's Qwen3-Coder, an open-source AI model, assists with software development and is particularly optimized for complex coding and autonomous programming tasks.
The Qwen3-Coder-480B-A35B-Instruct version features a 480-billion-parameter Mixture-of-Experts architecture, boasting 35 billion active parameters and a native context window of up to 1 million tokens.
Alibaba claims the model sets new performance benchmarks for agentic coding and is competitive with Anthropic's Claude Sonnet 4 in terms of capabilities.
The Qwen Code command-line tool, derived from Gemini Code, integrates seamlessly with popular developer tools for improved agentic coding experiences.
Collection
[
|
...
]