Do you feel that with each new generation of LLMs you actually get better results? For example, you don’t change how you use Claude Code/Cursor - you use the same prompts and the same tools, yet you get different results when you switch from, say, Sonnet to Fable? I personally don’t feel much di...