Pillar · Local AI
Run AI on your own hardware
Running a large language model at home comes down to VRAM above all else. I run this gear on my own bench, so here's which card to buy, which prebuilt box makes sense, and what any of it will actually fit.
Ranked
GPU comparison
The best GPUs for local AI, ranked by VRAM, bandwidth and value - consumer through workstation cards.
Open →Prebuilt
AI PCs
DGX Spark, Mac Studio, Strix Halo mini-PCs and prebuilt towers - the machines built to run big models.
Open →How-to
Guides
Run an LLM locally, understand Strix Halo, read your GPU with nvidia-smi and power-cap it for inference.
Open →Software
AI Computing
The serving stack, MCP tooling and agents - what runs on the hardware.
Tool
VRAM calculator
Pick a card and a model, see what fits before you spend.
Live
Top open models
The most-downloaded open LLMs, and what it takes to run them.
The other half
Bitcoin at home
Same GPUs, same bench - the mining side of the site.