← Back to gear list
🤖 AI & TechEssentialLocal AI Server (192.168.1.6)
GX10 / DGX Spark (Hermes AI)
My dedicated on-prem AI server — the GX10 runs Hermes models locally for the heavy background work that the M3 Mac Studio is too small for.
The NVIDIA DGX Spark (codename GX10) lives at 192.168.1.6 and serves as my dedicated local AI compute engine.
Powered by Hermes local AI models for ultra-fast inference response times, high context bandwidth, and zero cloud API dependency.
Handles continuous background tasks, automated data processing pipelines, and Hermes tool invocations with high memory throughput.
Reachable from the M3 Mac Studio over SSH — this is the rig that hosts the gear site you are reading right now.
The Good
- • Dedicated on-prem LLM inference with Hermes — no cloud API cost
- • High memory bandwidth for batched local model workloads
- • Powers the site / pipelines / data processing without round-tripping to a cloud provider
- • Complete data privacy with on-device execution
Watchouts
- • Higher power draw during continuous heavy inference batches
- • Requires dedicated Linux host management and server configuration
- • Not portable — lives at 192.168.1.6 in the rack
No commercial relationship. This is the rig that serves this very website.