Zenlayer 发布全球 AI 推断平台, 将悬浮率削减高达40%, 在300多个全球地点优化 GPU 使用 。 Zenlayer launches global AI inference platform, cutting latency by up to 40% with optimized GPU use across 300+ global locations.

Zenlayer推出了分布式推论,这是一个全球平台,通过优化GPU的使用,并通过其在50个国家的300+点存在网络减少延迟,简化了AI模型的部署。 Zenlayer has launched Distributed Inference, a global platform that simplifies AI model deployment by optimizing GPU use and reducing latency through its network of 300+ points of presence in 50 countries. 该系统使用先进的时间安排、路由和记忆管理,以便能够在边缘进行实时推断,将延迟率缩短40%,并支持使用自动化管弦和即时使用框架的多种模型。 The system uses advanced scheduling, routing, and memory management to enable real-time inference at the edge, cutting latency by up to 40% and supporting a wide range of models with automated orchestration and ready-to-use frameworks. 它消除了客户管理基础设施的需要,使得各区域能够更快、更具成本效益地扩大规模。 It eliminates the need for customers to manage infrastructure, allowing faster, more cost-effective scaling across regions. Zenlayer网络在25毫秒内覆盖全球85%的互联网用户, Zenlayer’s network reaches 85% of the global internet population within 25 milliseconds, marking a major step forward in delivering reliable, real-time AI intelligence worldwide.