Computers
Peripherals
Components
Supplies
Audio Visual
Mobile Tech
Software
Networking
My products
Product guidesCOLLECTION
Local AI
Running large language models on hardware you own: what it takes, what will actually fit, what to buy in a difficult market, and how to turn a model into a proper service on your own network. Start at the top if you are new to it, or jump straight to the hardware if you already know what you want.
Start here
Getting started with local AI
What local AI actually is, why you might want it, the one number that decides what you can run, and an honest account of what it is still bad at.
Software
Ollama: a beginner's guide
Installation, pulling models, what the tags mean, Modelfiles, and turning Ollama into a network service the rest of your homelab can call.
Hardware
How much VRAM do you actually need?
Model size, quantisation, the KV cache nobody budgets for, and why bandwidth decides how fast it feels once it fits.
Hardware
The best GPUs for local AI right now
NVIDIA, AMD, Intel Arc Pro, the used market and the datacentre-card trap — in the worst GPU market for years. Prices checked 28 August 2026.
Homelab
Running local AI in a homelab
Where the GPU should live, how to pass it through on Proxmox, what to run in front of it, and the security detail most guides skip over entirely.
Nothing matches that. Try “VRAM”, “Ollama” or “Proxmox”.