All Hardware for Running LLMs Locally

Choosing Hardware for Local AI

VRAM is the most important specification for running LLMs locally. Discrete GPUs offer dedicated VRAM, while Apple Silicon Macs share unified memory between CPU and GPU. AI development kits like NVIDIA Jetson provide optimized inference on the edge. Pick your hardware below to see which models you can run.

NVIDIA, AMD, and Intel GPUs — VRAM is the key spec for local LLMs

MacBooks and other laptops with unified or dedicated GPU memory

Mac Studio, Mac Mini, custom PCs, and workstations with maximum memory

NVIDIA Jetson, dedicated inference hardware, and edge AI devices

iPhones and Android phones — on-device LLM inference with no internet required

iPads and Android tablets for on-device AI with fixed memory pools

Enterprise servers with 80–640 GB VRAM for running frontier-scale models

Browse by VRAM

Find the best models for your VRAM tier