"llama" is the name of the open-weights model released by meta for which llama.cpp (not by meta) was originally designed. llama.cpp now supports many models, not just llama. they're separate projects with separate authors.

ah, OK, thanks for the clarification