On August 13, Alibaba opened up the large-scale MoE model Qwen3.8-2.4T-A95B, and the Zhongzhi FlagOS community simultaneously completed multi-chip adaptation. Currently, Qwen3.8-2.4T-A95B has completed multi-core adaptation, precision alignment, and deployment verification based on the FlagOS unified open source technology stack on 9 AI chips, including Pingtou, Nvidia, Moore Thread, Huawei Shengteng, Mu Xi, Kunlun Chip, Haiguang, Qingwei Intelligence, and Suiyuan.

On August 13, Alibaba opened up the large-scale MoE model Qwen3.8-2.4T-A95B, and the Zhongzhi FlagOS community simultaneously completed multi-chip adaptation. Currently, Qwen3.8-2.4T-A95B has completed multi-core adaptation, precision alignment, and deployment verification based on the FlagOS unified open source technology stack on 9 AI chips, including Pingtou, Nvidia, Moore Thread, Huawei Shengteng, Mu Xi, Kunlun Chip, Haiguang, Qingwei Intelligence, and Suiyuan.