Skip to main content
ModelTerms

Prompting · beginner

Zero-Shot (zero-shot prompting)

Zero-shot prompting asks the model to perform a task without showing any examples — only the instruction and the input. Modern instruction-tuned models do this well.

Explanation

"Translate this to French: Hello" is a zero-shot prompt. No examples of English-to-French pairs are included; the model is expected to know how to translate from instruction alone.

Before instruction tuning, zero-shot performance was weak and few-shot was essential. After instruction tuning and RLHF, most everyday tasks (summarize, translate, classify, extract) work fine zero-shot. Few-shot examples are still useful for unusual formats, niche domains, or very specific style requirements.

Examples

  • "Summarize this article in 2 sentences."
  • "Classify this email as spam or not."

Frequently asked

What is Zero-Shot?

Zero-shot prompting asks the model to perform a task without showing any examples — only the instruction and the input. Modern instruction-tuned models do this well.

What is an example of zero-shot?

"Summarize this article in 2 sentences."

How is Zero-Shot related to Few-Shot?

Zero-Shot and Few-Shot are both prompting concepts. Few-shot prompting includes a small number of input-output examples directly in the prompt so the model can pattern-match without any fine-tuning.

Is Zero-Shot considered beginner?

Zero-Shot is generally considered beginner-level material in the AI and LLM space.

Side-by-side comparisons

Sources