---
格式版本: 2
标题: "H100 Multi-Node Serving Now Available for Model Import in OCI Generative AI"
原文链接: "https://docs.oracle.com/en-us/iaas/releasenotes/generative-ai/h100-multinode-imported-models.htm"
发布日期: "2026-07-15"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "manual"
发布时间证据: "正文中标题附近明确标注的发布日期，符合发布时间定义，且优先级最高。"
发布时间校准原因: "正文中标题附近明确标注的发布日期，符合发布时间定义，且优先级最高。"
发布时间校准置信度: "manual-confirmed"
发布时间候选数量: 0
发布时间严格候选数量: 0
发布时间原页读取状态: "manual"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-17T06:40:22+08:00"
发布时间仲裁状态: "manual"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-16T21:25:50+08:00"
入库时间: "2026-08-16T13:32:05.844Z"
来源平台: "固定入口"
搜索渠道: "fixed_url"
搜索词: "https://docs.oracle.com/en-us/iaas/releasenotes"
匹配关键词:
  - "AI"
  - "deployment"
相关厂家:
  - "Oracle"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 39
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "正文仅提及OCI支持H100多节点部署，未涉及超节点/AI Rack/机柜级架构、供电、散热、互连等核心主题，技术细节极少，仅属云服务功能更新，不满足项目关注范围。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-16T21:32:11+08:00"
AI主题相关性: 4
AI来源权威性: 15
AI新颖性: 5
AI技术细节: 2
AI商业部署信号: 5
AI完整性: 8
采集批次: "2026年8月16日20点37分10秒"
采集批次ID: "20260816-203710-016"
去重键: "https://docs.oracle.com/en-us/iaas/releasenotes/generative-ai/h100-multinode-imported-models.htm"
---

H100 multi-node serving is now available for imported models in OCI Generative AI.

You can now deploy [deepseek-ai/DeepSeek-V4-Pro](https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro) and [zai-org/GLM-5.2-FP8](https://huggingface.co/zai-org/GLM-5.2-FP8) using the `H100_X16` shape. This shape provides 16 H100 GPUs across two nodes, with 8 GPUs per node.

For available hardware and deployment steps, see [Dedicated AI Cluster Unit Sizes for Imported Models](https://docs.oracle.com/iaas/Content/generative-ai/manage-imported-models.htm#unit-sizes). For the complete list of models compatible for import, see [Compatible Models for Import](https://docs.oracle.com/iaas/Content/generative-ai/imported-models.htm).