TL;DR: On a low-spec laptop, most Ollama slowdowns are memory pressure. Start with ollama ps to see whether inference is on GPU or CPU and what context is active. If memory is tight, prefer a Q4 quantized variant of the…