Alibaba launches Qwen3.8-Max, its largest model yet

Alibaba officially launched Qwen3.8-Max on August 3, 2026, a 2.4-trillion-parameter mixture-of-experts model (95B active) with a 1M-token context window, available now via Alibaba Cloud's Model Studio. It's also the first Qwen-Max-class model Alibaba is open-sourcing, with weights following the week after launch.

Why it matters

This is Alibaba's largest model to date and its first flagship-class model to ship with open weights, not just an API — a meaningfully different move from the closed-weight releases most frontier labs have stuck to this year.

Who should care

Developers evaluating open-weight models for long-context or large-codebase tasks

What you can do

Qwen3.8-Max is live now on Alibaba Cloud's Model Studio; open weights follow roughly a week after launch for local or self-hosted use.