To be clear, this is not comparable directly to llama.cpp -- Gemma models work on llama.cpp and we encourage people who love llama.cpp to use them there. We're

To be clear, this is not comparable directly to llama.cpp -- Gemma models work on llama.cpp and we encourage people who love llama.cpp to use them there. We're also launched with Ollama. Gemma.cpp is a highly optimized and lightweight system. The performance is pretty incredible on CPU, give it a try =)