What Is Fine-Tuning?

Fine-tuning trains a model on your examples to change its behavior. Here's what it is, when it helps, and why to try prompting and RAG first.

What fine-tuning is

Fine-tuning further trains a pre-trained model on your own examples to change its default behavior—teaching a consistent style, format, or specialized task. See AI model fine-tuning.

What it costs

Requirement	Impact
Quality training data	Effort to prepare
Compute	More than prompting
Maintenance	Redo as base models improve

Fine-tuning produces a customized model, but at real cost.

When it's worth it

Fine-tune when you need consistent style/format, a specialized task, or shorter prompts at scale—and prompting plus retrieval isn't enough.

Why try prompting and RAG first

For most needs, prompting plus RAG achieves the goal more cheaply and flexibly—and RAG adds live knowledge fine-tuning can't. See fine-tuning vs RAG. They solve different problems and are sometimes combined.

The common mistake

Reaching for fine-tuning first—it's costlier and slower than the alternatives for most cases. Exhaust prompting and retrieval before committing.

Why FISTA

FISTA Solutions uses the cheapest approach that works—prompting and RAG first, fine-tuning only when it pays—through AI enablement, backed by 150+ projects across 12+ countries.

Deciding whether to fine-tune? Talk to FISTA.