Fine-tune or prompt

Fine-tuning a model sounds professional. It also means data to prepare, training to maintain and a cost that comes back with every change. Before going there, we make sure we actually need it.

What prompting already does very well

Output format, tone, business rules, examples: the vast majority of needs are handled by good instructions and a few well-chosen examples. It is immediate, adjustable, and it creates no debt.

When fine-tuning is worth it

• A very specific style or format, required at very high volume.
• A latency or a cost to compress on a task repeated millions of times.
• A domain where prompting hits a ceiling despite every effort.

Our rule: prompt first, measure, and only fine-tune if the numbers demand it. Nine times out of ten, they don't.