Alibaba debuts Qwen3.8-Flash: 1-ninth training cost, 1 million token context
Alibaba just rolled out Qwen3.8-Flash, its latest AI model designed to handle everything from coding help to big office tasks.
The standout? It's way more affordable to train, costing only about one-ninth of the previous version, and can process huge files thanks to an expandable context window (up to 1 million tokens for those deep research dives).
API fees ¥1/¥3, weights open-sourced
Qwen3.8-Flash is easy on the wallet, with API access 1 yuan ($0.1488) per million input tokens and 3 yuan per million output tokens.
Plus, Alibaba has open-sourced the weights for Qwen3.8-Flash-Next, giving developers a sneak peek at what's coming in the next-gen Qwen4 family.
Alibaba holds HK$80 billion share sale
With e-commerce growth slowing down, Alibaba is doubling down on AI innovation, including an HK$80 billion share sale on Sunday to fund these efforts, as competition heats up in China's tech scene.