My theory: improving LLM output always requires a higher token count and, therefore, cost.

Run your business on a poem generator, I dare you!