Alibaba released three new AI models from its Qwen3.8 lineup, led by Qwen3.8-Max with 2.4 trillion total parameters, of which 95 billion are active at any given time using a mixture-of-experts architecture. The Qwen3.8-27B model, designed to run on consumer hardware, crossed the 1 million download milestone shortly after launch. Alibaba also introduced Qwen3.8-Flash with 125 billion parameters as a cost-effective option. All three models are available as open-weight on Hugging Face and ModelScope, with Qwen3.8-27B released under the Apache 2.0 license, enabling unrestricted commercial use. This initiative is part of a broader strategy that has already produced over 460 models, spawned more than 300,000 derivative models, and accumulated over 3 billion downloads globally. Alibaba’s stock price rose following the announcements.
Source: Read the original article
