AI 101

Zero-shot prompting

Also known as: zero-shot

Inference

Zero-shot prompting is asking a model to perform a task with no examples in the prompt — just the instruction. It has to draw entirely on what it learned in pretraining.

  • “Translate this to French.” (no examples) → zero-shot.
  • “Translate this to French:” + 3 examples → few-shot.

It’s the simplest form, and modern models handle a lot zero-shot.

When it's enough: for common, well-learned tasks, zero-shot works. When output quality or format slips, adding 2–3 examples (few-shot) is the cheapest fix.