PulseAugur
EN
LIVE 10:07:36
中文(ZH) 阿里云：真武芯片超节点已成功适配Qwen3.8，上线百炼提供推理服务

Alibaba Cloud's Zhenwu chip super node runs Qwen3.8 trillion-parameter model

By PulseAugur Editorial · [3 sources] ·

Alibaba Cloud has successfully adapted its Qwen3.8 model, a 2.4 trillion parameter large language model, to run on its Zhenwu M890 super node. This integration, available on Alibaba Cloud's Bailian platform, marks the first time a super node has supported a model of this scale, offering enhanced inference services. The Zhenwu M890 chip and its super node instance, featuring high-speed interconnection and substantial memory, are optimized for large-scale MoE models, promising improved inference efficiency and reduced costs. AI

IMPACT Enables efficient inference for extremely large models, potentially lowering costs and accelerating adoption of advanced AI capabilities.

RANK_REASON Infrastructure and model integration news from a major cloud provider.

Read on 量子位 (QbitAI) →

AI-generated summary · Google Gemini · from 3 sources. How we write summaries →

Alibaba Cloud's Zhenwu chip super node runs Qwen3.8 trillion-parameter model

COVERAGE [3]

  1. 量子位 (QbitAI) TIER_1 中文(ZH) · 量子位的朋友们 ·

    Alibaba Cloud: Zhenwu chip supernode has successfully adapted Qwen3.8, providing inference services on Bailian.

  2. 36氪 (36Kr) TIER_1 中文(ZH) ·

    Alibaba Cloud: Zhenwu chip supernode has successfully adapted Qwen3.8, providing inference services on Bailian.

    36氪获悉，7月23日，阿里真武M890超节点已成功适配Qwen3.8，并上线阿里云百炼平台提供模型推理服务，这是国内首个成功运行超2万亿参数大模型的超节点。

  3. 雷峰网 (Leiphone) TIER_1 中文(ZH) ·

    Alibaba Cloud: Zhenwu chip supernode has successfully adapted Qwen3.8, providing inference services on Bailian.

    <p>7月23日消息，阿里真武M890超节点已成功适配Qwen3.8，并上线阿里云百炼平台提供模型推理服务，这是国内首个成功运行超2万亿参数大模型的超节点。</p><p>&nbsp;</p><p>Qwen3.8是阿里最新旗舰模型，参数规模高达2.4万亿，跑通这一规模的模型需要将参数分散到几十张甚至上百张高速互联的GPU卡，让万亿参数“跑得又稳又快”。然而，普通AI集群因通信带宽的局限，无法满足高吞吐、低延迟、高并发的需求。</p><p>&nbsp;</p><p>真武M890是平头哥新一代训推一体AI芯片，原生支持FP32到FP4等多种数据精度，可应用于高…

RELATED ENTITIES

RELATED TOPICS