Alibaba's Qwen to open-source Qwen3.8-Flash-Next model

Alibaba's AI team, Qwen, is scheduled to open-source its Qwen3.8-Flash-Next model and an FP8 version on August 26 at 11 p.m. Beijing time. This model is described as a multimodal mixture-of-experts system built upon the next-generation Qwen4 architecture.

The team stated that the early release is intended to provide the developer community with an opportunity to prepare for the full Qwen4 model family. Further details regarding the model's specifications or performance have not yet been disclosed.

Alibaba's move towards open-sourcing AI models signals a strategy to foster broader community development and potentially accelerate innovation within the rapidly evolving AI landscape. This approach can lead to wider adoption and diversification of AI applications.

The release comes at a time of intense competition and rapid advancement in the field of large language models, with various entities vying for influence and contributions within the AI ecosystem.