Alibaba Drops Qwen3.8-Max AI Model to Automate Coding Entirely
Alibaba just launched a giant multimodal AI agent designed to code, debug, and manage complex software projects for days without human intervention, all while slashing market prices.
The new Qwen3.8-Max flagship model from Alibaba packs 2.4 trillion total parameters, running on a Mixture of Experts architecture that only fires up 95 billion parameters per query to keep compute costs ridiculously low.
Capable of processing up to a 1-million token context window across text, image, and video, the model weaves visual reasoning directly into its step-by-step logic loop rather than treating images as static snapshots.
In a ten-day test run on GitHub, the AI agent built an entire software CLI tool from a clean directory, autonomously planning architecture, executing tools, running tests, and fixing its own bugs in an open log.
To undercut western competitors, access costs just $2 per million input tokens and $6 per output million, with open weights scheduled for public release next week alongside a smaller 27-billion variant.
Watching a Chinese cloud giant release a frontier-class, open-weight agentic model at budget-bin pricing leaves legacy software engineers wondering if updating their resumes is even worth the effort anymore.
Source: Qwen Blog
Comments
Help shape the next version: Add context or suggest a correction. AI review can add points toward a rewrite. Reviews and updates may take time; a full meter does not guarantee a new version.