Fine-tuning is the process of adjusting a pre-trained model's weights on a smaller, task-specific dataset. Unlike training from scratch (millions of dollars), fine-tuning is cheap and fast — typically hours to days on a few GPUs. Use cases: matching a brand's voice in copy, classifying support tickets, generating code in a private framework, recognizing legal jargon. The trade-off: every fine-tune locks you to a specific model version. When OpenAI releases GPT-5, your GPT-4 fine-tune is stuck unless you re-fine-tune. By 2026, most teams prefer RAG or LoRA fine-tuning (cheaper, swappable) over full fine-tuning, except for high-volume, narrow tasks where the latency and cost savings of a fine-tuned model outweigh the maintenance burden.
GLOSSARY
What is Fine-tuning?
Taking a pre-trained AI model and continuing to train it on your specific data so it specializes for your use case (medical, legal, customer support style, etc.).
RELATED TERMS
LoRA (Low-Rank Adaptation)
An efficient fine-tuning technique that trains a small 'adapter' on top of a frozen base model — fast to train, tiny to store, and stackable.
RAG (Retrieval-Augmented Generation)
A technique that lets an LLM answer questions about information it wasn't trained on — by retrieving relevant documents at query time and stuffing them into the prompt.
LLM (Large Language Model)
An AI system trained on massive text datasets to predict and generate human-like text — the technology behind ChatGPT, Claude, Gemini, and most modern AI chatbots.
Back to the AI Glossary