llama.cpp

Georgi Gerganov's CPU-and-Metal LLM inference engine in C/C++. Powers Ollama, LM Studio, and basically every 'local LLM' app on the planet.

Open source ↗