Skip to content

Alibaba's Qwen4-Max leaves preview, pricing finally goes public

The flagship model moves off its Token Plan-only preview and onto pay-as-you-go pricing across six regions, with native vision and video input and a 1M-token context window.

Source: Alibaba Cloud

Alibaba took Qwen4-Max out of preview and into general availability on 16 November, five days after the Token Plan-only preview opened. Alibaba Cloud’s Model Studio docs now list it as the latest model, with pricing attached for the first time.

The context window stays at 1 million tokens, with a maximum input of 991,808 tokens and output capped at 131,072. The combined budget for thinking plus final answer is 262,144 tokens, down from the preview’s 393,216. Input now accepts text, image and video; output is text only. It’s live in six regions, including Singapore, Frankfurt, Tokyo and US (Virginia). Pricing is $1.50 per million input tokens, $4.50 per million output tokens, and $0.075 per million for cached input reads. Alibaba says the mixture-of-experts model, 2.4 trillion parameters total with 71 billion active per token, matches its predecessor Qwen3.8-Max’s output using roughly half the tokens, and cuts its own measured cost per task by more than 60%.

Why it matters: the preview shipped with no pay-as-you-go access and no stated replacement date. Real, public per-token pricing and multi-region availability is what turns a model from “worth a look” into something you can actually budget against and benchmark for production use.

The caveat: there are still no open weights. This is API-only, and the thinking-plus-answer token budget actually shrank from the preview, so anything tuned against that limit needs retesting.

Share

Related reading

More from The Wire

All briefs