Fine-tuning
Fine-tuning trains a model on your data. It can cut prompt size and calls, but has upfront training cost to weigh.
Definition
Fine-tuning adapts a base model to your tasks using example data, so it needs less instruction and few-shot context.
Why it matters for cost
A fine-tuned model can shrink prompts and reduce the need for retrieval or multi-step prompting—lowering per-call tokens. But it adds training cost and maintenance.
Practical tip
Fine-tune only after you've measured that prompt shrinkage outweighs training and upkeep cost.
Related reading
Batch APIs process large async request sets at a discount. Ideal for jobs that can tolerate delayed results.
The context window is the maximum tokens a model can see in one request. Bigger windows cost more per call when filled.
Embeddings turn text into vectors for search and retrieval. They're usually far cheaper per token than generation.
Distillation trains a small model to mimic a large one, trading some quality for much lower per-token cost at scale.