Alibaba Releases Qwen3.8-Max: 2.4T Parameters Built for Days of Autonomous Work
Alibaba has announced Qwen3.8-Max with 2.4 trillion parameters: in 16 days the model wrote a CLI tool with 265 commits and outperformed 458 out of 526 teams in an e-commerce simulation. Weights drop on Hugging Face next week.
Alibaba has released Qwen3.8-Max — 2.4 trillion parameters, 95 billion active per request, architecture built for autonomous operation over days. In 16 days the model independently wrote a CLI tool with 265 commits, improved the results of a research paper by 2.7 points, and in an e-commerce simulation outperformed 458 out of 526 human teams, growing capital by 316%.
Weights will be released on Hugging Face and ModelScope next week — a frontier model for long-horizon agentic tasks going fully open. The race is shifting toward autonomy: not "what can the model do in one prompt," but "how many days can it run without human involvement."
Source: the-decoder.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →▌ Related guides
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles