Comment by jonahbenton

Run llama.cpp on your machine and point opencode to it