llama.cpp
Explains llama.cpp as the local inference base layer, showing how GGUF and quantisation support cross-platform runs, and sorts out which jobs it takes on rather than a ready-to-use local tool.
llama.cpp is the open-source engine at the base of the local LLM ecosystem, running quantised models on Mac, Linux, Windows, servers and edge devices. It supports local formats such as GGUF and prioritises low-resource inference and cross-platform execution, so many one-click local tools build on it. It is closer to an engine than a finished car: command line, model format and hardware acceleration stay yours to tune.
