# 推理专题日跟踪 · 2026-09-10 > 日期:2026-09-10(Asia/Shanghai)。专题正文:`intel/topics/inference.md`。原料:`archive/2026-09-10/`。 ## 摘要 - **新稿 · RH AI 3.5:** priority-aware serving(GA)、vLLM CPU/NVMe KV offload(GA)、Inference-time scaling(GA)、llm-d 分布式推理扩展至 CoreWeave CKS / Azure AKS(GA)与 EKS(TP)。 - **新稿 · Rafay×RH:** Token Factory 叠在 vLLM/llm-d 上做多租户计量推理云。 - **续档:** VCF×Cisco AI fabric(推理/RAG 南北向);RH context wall(09-09)。 ## 原文摘录 ### RH AI 3.5 · 推理控制面 - 归档:`redhat/red-hat-ai-3-5-scaling-and-governing-ai-agents-in-production-0f281ed9.txt` - URL:https://www.redhat.com/en/blog/red-hat-ai-35-scaling-and-governing-ai-agents-production > … priority-aware serving (GA)… vLLM introduces CPU offloading (GA) and NVMe storage offload for the KV cache… > Inference-time scaling (GA)… adaptive self-consistency… > … distributed inference reaches GA on CoreWeave CKS and Microsoft Azure AKS, while Amazon EKS joins in technology preview. ### Rafay×RH Token Factory - 归档:`redhat/rafay-and-red-hat-publish-joint-reference-architecture-for-sovereign-a-a699b154.txt` > … Rafay Token Factory adds a metered, multi-tenant billing and API layer on top of Red Hat AI's vLLM and llm-d capabilities… ## 检索范围 - archive/2026-09-10/redhat、rafay、vmware;对照 2026-09-09 inference 日跟踪