Fine-tuning is the process of taking a pre-trained AI model and continuing to train it on a smaller, specialized dataset so it learns a particular style, domain, or behavior. Think of it like hiring someone with a general education and then giving them a targeted apprenticeship.

During fine-tuning, the model's existing weights — the billions of numerical values that encode its knowledge — are adjusted slightly based on the new examples. The model doesn't start from scratch; it builds on what it already knows. This makes fine-tuning much cheaper and faster than training from zero.

Fine-tuning is used for all kinds of goals: making a model write in a specific tone, follow a company's formatting rules, answer only about a narrow topic, or even mimic a vintage writing style. It's one of the most practical tools developers have for customizing general-purpose AI into something purpose-built.