Definition
Zero-shot prompting asks a language model to perform a task using only the instruction itself, without any examples. Modern frontier models (GPT-5, Claude Opus 4, Gemini 2.5) handle zero-shot well for most general tasks. Few-shot wins when the task has nuance, custom format, or specific style requirements. Zero-shot is faster to compose; few-shot is more accurate when accuracy matters.
Example
'Classify this review as positive, negative, or neutral: "Decent product but shipping was slow."' — no examples provided, model classifies directly.
When to use
Simple classifications, well-known tasks, when speed beats precision, exploratory work.
Also known as
zero shot