NVIDIA IFA 2026加速本地AI：RTX Spark与多模型更新
原标题：Sparks Fly: NVIDIA Accelerates Local AI at IFA 2026
AI 摘要
在IFA 2026上，NVIDIA联合微软及合作伙伴宣布加速本地AI推理，推出RTX Spark迷你PC，并简化Hermes Agent、OpenClaw和Perplexity Portable Computer等应用的本地模型设置。同时发布了多项模型更新，包括Nemotron 3.5 Lightning、GLM-5.3-Flash、Qwen3.8系列等，并推出NVIDIA PAIR工具以优化本地网络推理分发。
正文节选
Frontier intelligence is going local. At IFA 2026, NVIDIA, Microsoft and its partners are teaming up to provide faster inference and new tools that make agents easier to set up and run locally on NVIDIA hardware. New compact NVIDIA RTX Spark Windows PCs are also coming in October to give AI enthusiasts, developers and creators more ways to run capable agents locally and securely. Today’s announcements include: - Simplified local AI support for NVIDIA GPUs is coming in Hermes Agent, OpenClaw and