llama.cpp
Short answer
llama.cpp is the open-source C++ inference engine that most local AI tools are built on, including Nightjar.
It is the reason local AI became practical on ordinary hardware: it runs quantised models on CPU and GPU, on every major platform, with no Python environment and no framework install.
Almost anything you read about GGUF files, quantisation levels or GPU layer offloading is describing llama.cpp's behaviour, which is why those concepts transfer between tools.
Related terms
Other ways people ask this
- what is llama cpp