ThinkFacility

Alibaba API pricing

What the Qwen models cost per million tokens, straight from Alibaba's own docs. Put your own usage in the boxes and the table works out the monthly bill.

3 priced models · updated 2026-09-05 12:15 UTC

What would this cost me?

One request, and how often you send it. We count words and turn them into tokens at the usual English rate.

Cheapest for that workload: Qwen3.8-Flash at $0.27 a month.

Model Lab Input Output Context Your month
Qwen3.8-FlashAlibabaAlibaba$0.15$0.471M$0.27
Qwen3.7-PlusAlibabaAlibaba$0.40$1.601M$0.86
Qwen3.8-MaxAlibabaAlibaba$2.00$6.001M$3.52

The short version

Qwen3.8-Flash is the cheapest Alibaba model you can pay for, at about $0.27 a month for steady daily use. Qwen3.8-Max does the same work for $3.52.

Alibaba prices its hosted models through Model Studio and publishes the open weights separately.

No published price

Qwen3.8-27B has no price in the docs we read, so it is left out rather than guessed at.

Everyone else

The full table puts Alibaba next to every other lab on one page, and the comparison tool takes any two models side by side.