Ollama, Docker Model Runner, LM Studio, and llama.cpp workflows commonly use GGUF-style artifacts. Q4_K_M is one of the most popular practical defaults: small enough to fit, usually good enough for ...