What I really want for coding is a terminail that runs a local model to churn out code as fast as possible that then i can use to ask gpt-max-thinking mode ”find all the major issues”, until it converges. Anyone have a setup like this?(toly 🇺🇸)