What is Fine-tuning?
Fine-tuning further trains a model on your own examples so its default behaviour changes — useful for a fixed style or format at scale, unnecessary for most tasks.
Prompting changes what a model does on one request; fine-tuning changes what it does by default, by training it on hundreds or thousands of example input → output pairs. It's how vendors turn base models into chat models, and how companies get a consistent brand voice or a niche format without a long prompt every time.
It's rarely the first tool to reach for. Good prompts with examples get most of the benefit at zero cost; RAG handles "the model doesn't know my data" better than fine-tuning does. Fine-tune when the prompt is huge, the task is repetitive and the volume is high.
- Try few-shot prompting and RAG first.
- Fine-tune for style and format, not for facts.
- Keep an evaluation set so you can prove it helped.
Use it right now
Ask our brain anything on the homepage — it remembers the whole conversation — or write a brief in the Studio and see the prompt it compiles to.