阿里开源Qwen3.8-2.4T-A95B模型权重
Alibaba Opens Qwen3.8-2.4T-A95B Weights as First Qwen-Max-Class MoE
精选理由
阿里开源了首个Qwen-Max级MoE模型，参数规模达2.4万亿，专为智能体任务优化。
阿里Qwen团队发布了Qwen3.8-2.4T-A95B模型的开放权重，这是一个拥有2.4万亿参数的混合专家模型(MoE)，其中活跃参数为950亿。该模型采用混合注意力机制和长上下文设计，专门针对智能体工作负载优化。同期还提供了SageMaker HyperPod和vLLM的部署方案。
原文 · pandaily
Alibaba Opens Qwen3.8-2.4T-A95B Weights as First Qwen-Max-Class MoE
Alibaba's Qwen team published open weights for Qwen3.8-2.4T-A95B, a 2.4-trillion-parameter MoE with 95 billion active parameters. Hybrid attention and long context target agentic workloads, with same-period SageMaker HyperPod and vLLM recipes.
