---
title: Zero-Shot Prompting — Prompt Engineering Technique
description: Zero-shot prompting asks a model to perform a task from an instruction alone, with no worked examples. It relies entirely on knowledge the model already learned during pre-training and instruction tuning.
canonical: https://prompeteer.ai/techniques/zero-shot-prompting
category: Foundational patterns
---

# Zero-Shot Prompting

Zero-shot prompting asks a model to perform a task from an instruction alone, with no worked examples. It relies entirely on knowledge the model already learned during pre-training and instruction tuning.

## What it is

Zero-shot prompting is the simplest way to use a large language model: you describe the task in plain language and let the model respond without showing it any solved examples. Instruction-tuned models are trained to follow such directives directly, so for common, well-understood tasks a clear instruction is often all that is needed.

Because there are no examples to anchor the output, zero-shot performance depends heavily on how precisely the instruction is phrased and how familiar the task is to the model. It is the natural baseline to try before investing in few-shot examples or more elaborate scaffolding.

## When to use

- The task is common and unambiguous (summarize, translate, classify into obvious categories).
- You want the cheapest, lowest-latency baseline before adding examples.
- Prompt length or token budget is tight and every example is costly.
- You are testing whether the model already "knows" the task before engineering further.

## How it works

1. State the task as a direct, specific instruction.
2. Add any constraints (tone, length, format) the output must satisfy.
3. Provide the input to operate on, clearly delimited from the instruction.
4. Read the result and, if quality is insufficient, escalate to few-shot or chain-of-thought.

## Illustrative structure

The structure is: [role or task instruction] + [explicit constraints] + [delimited input] → response. For example: "Classify the sentiment of the following review as positive, negative, or neutral. Review: <text>." No solved examples are included.

## Pitfalls

- Ambiguous instructions leave the model to guess format and scope, producing inconsistent output.
- Novel or domain-specific tasks the model has not internalized often need examples.
- Harder reasoning tasks may need chain-of-thought; a bare zero-shot instruction can skip the reasoning.

## Sources

- [Brown et al. (2020), Language Models are Few-Shot Learners](https://arxiv.org/abs/2005.14165)
- [OpenAI — Prompt engineering guide](https://platform.openai.com/docs/guides/prompt-engineering)

## Related techniques

- [few-shot-prompting](https://prompeteer.ai/techniques/few-shot-prompting)
- [chain-of-thought](https://prompeteer.ai/techniques/chain-of-thought)
- [system-prompts](https://prompeteer.ai/techniques/system-prompts)