Machine Learning
When to Fine-Tune a Language Model (and When Not To)
A decision framework for choosing between prompting, retrieval and fine-tuning.
Sense Central Editorial·September 3, 2026· 9 min read
Fine-tuning is often the first tool teams reach for when a model does not behave as expected. It is rarely the right one. Prompting improvements, better retrieval and smaller task-specific models solve most problems faster and cheaper.\n\nConsider fine-tuning when three conditions hold: the task is well-defined and repeatable, prompt engineering has plateaued, and you have hundreds to thousands of high-quality examples. Otherwise, invest in retrieval quality and prompt design first.
About the author
Sense Central EditorialThe Sense Central editorial team writes original guides across markets, technology, business and personal finance.