Build guide
Single RTX 4080 SUPER build for running 7B–14B models locally with llama.cpp or Ollama.
Budget
$2,800
Profile
LOCAL DEV
Target model
Llama 3.1 8B
52tok/s
44.2–59.8 tok/s decode on Llama 3.1 8B
Value: 18.57 tok/s per $1k
GPU
NVIDIA GeForce RTX 4080 SUPER
CPU
AMD Ryzen 7 7800X3D
Motherboard
MSI MAG B650 TOMAHAWK WIFI
RAM
Kingston FURY Beast 32GB (2x16GB) DDR5-5600
Storage
WD Black SN850X 1TB NVMe
PSU
Seasonic FOCUS GX-850 850W 80+ Gold
Case
Lian Li Lancool 216
Cooler
NZXT Kraken X73 360mm AIO