Alibaba Is Rolling Out Its ‘Most Agentic Code Model to Date’

Alibaba’s Qwen team has launched Qwen3-Coder, which it calls its “most agentic code model to date.” While it will be made available in multiple sizes, the most powerful variant — Qwen3-Coder-480B-A35B-Instruct — is being released first. The 480 billion parameter mixture-of-experts model has 35 billion active parameters supporting a context length of 256,000 tokens natively and 1 million tokens with extrapolation methods for “exceptional performance in both coding and agentic tasks,” explains the group, which claims the quasi-open source model has agentic coding, agentic browser use, and agentic tool use comparable to Anthropic’s proprietary Claude Sonnet 4.

Alibaba is making the Qwen3-Coder weights available from both Hugging Face and GitHub. The company is also providing a command-line interface (CLI) tool called Qwen Code for agentic coding. Forked from Gemini Code Assist, “Qwen Code has been adapted with customized prompts and function calling protocols,” the Qwen team notes in a blog post.

“While Qwen Code natively supports Qwen 3 Coder AI model, it can also be integrated with the OpenAI software development kit when calling LLMs,” reports Gadgets 360, noting that the Qwen3-Coder model “can also be used with Claude Code, but developers will need to request an API key on Alibaba Cloud Model Studio platform.”

Since Alibaba is not providing the actual training scripts, the model can be used as is but not modified. It is made available “with a permissive Apache 2.0 license which allows for both academic and commercial usage,” Gadgets 360 says.

Computerworld talks up China’s open approach to AI models, quoting Everest Group Director Abhishek Sengupta saying that “as the U.S. increasingly takes a restrictive approach towards sharing its AI ecosystem, putting curbs on chip exports, it may very well leave the [global] market open to Chinese alternatives.”

Lian Jye Su, a chief analyst at Omdia, tells Computerworld that “the adoption of foreign AI models is becoming rarer in both the U.S. and China due to many reasons, including those like regulatory, trust, and national security concerns.” Su says he expects the rise of sovereign AI “to become a major trend in mission-critical and business-critical applications that prefer to have GenAI models being supported by local infrastructure and cloud services.”

Alibaba shares Qwen 3 Coder benchmark test results against models including Deepseek-V3 and GPT-4o.

VentureBeat points out that the latest Qwen family updates are “already attracting attention from AI power users in the West for their top performance, in one case edging out the new Kimi-2 model from rival Chinese AI startup Moonshot, released in mid-July.”

Related:
Alibaba’s New Qwen3 Reasoning Model Tops OpenAI and Google Benchmarks in Major Open-Source Release, WinBuzzer, 7/27/25
Alibaba’s Qwen3 Upgrade Beats OpenAI and DeepSeek on Benchmarks, Digital Watch, 7/23/25

No Comments Yet

You can be the first to comment!

Leave a comment

You must be logged in to post a comment.