Glossary
An existing model is further trained on your own examples, to hit style, format or domain language reliably.
Fine-tuning changes the behaviour of a model, not its knowledge in any useful sense. For current facts from your own data, retrieval is the right tool, not retraining.
It pays off when format and tone have to be exact, when the task is narrow and frequent, or when a smaller model suffices once it knows the job.