Running Hermes AI Locally on the Asus P16 RTX Spark (128GB Unified Memory)

Nov 15, 202642 views4:21

In this walkthrough, we’ll see what it’s like to run Hermes AI locally on an Asus P16 with the RTX Spark and 128GB of unified memory, using a local Qwen 3.8 Flash Next model. We’ll compare it with Copilot, try out research and terminal workflows in Windows PowerShell, and even point a MacBook Air at the local gateway to offload heavier AI tasks over the network.

Open in YouTube

Description

Curious what running local AI models actually looks like in a day-to-day workflow? In this first look, I put the Hermes AI agent software through its paces on an Asus P16 equipped with the RTX Spark and 128GB of unified memory. Instead of relying purely on cloud subscriptions, I test how Hermes handles real research, terminal tasks, and network-wide access using a local Qwen 3.8 Flash Next model. If you're exploring local LLMs, AI agent harnesses, or high-memory mobile workstations, this walkthrough will help you see where local AI shines—and where cloud tools like Copilot still have the edge. What we cover in this video: - Hardware & Model Setup: Why 128GB of unified memory makes the Asus P16 RTX Spark ideal for local AI - Local AI vs. Copilot: Testing Hermes' web-search and research capabilities side-by-side against Microsoft Copilot on a data platform comparison prompt. - Command-Line Workflow: Running local AI prompts directly inside Windows PowerShell without leaving the terminal. - Personal AI Server Setup: Pointing the Hermes Mac app on a MacBook Air to the RTX Spark's local gateway so a lighter laptop can offload heavy AI processing over the local network. Subscribe to follow along as I build out more complex AI agent workflows, custom skills, and tool integrations with Hermes in upcoming videos. Let me know in the comments what local AI use cases you want to see tested next! 0:00 Introduction 0:28 Hermes Windows GUI 1:44 Copilot comparison 2:17 Hermes Powershell 2:56 Macbook Air via Gateway 3:33 Summary