llama.cpp

Port of Facebook's LLaMA model in C/C++ for local inference — It is the foundational engine that powers Ollama and most modern local LLM runners under the hood.