didn't Ollama vendor Llama cpp too?

Most projects typically start with llama.cpp and then move away to proprietary kernels