llama.cpp

Overview

Enables efficient local inference for models like Llama with M1/M2/M3 optimizations and broad hardware support. Essential for developers building privacy-focused AI apps without cloud dependency. Highly flexible for custom integrations.