---
title: "Meta Prompting"
source: "https://systems-analysis.info/eng/Meta_Prompting"
wiki: "systems-analysis.info/eng"
article: "Meta_Prompting"
language: "en"
categories:
  - "Category:English"
  - "Category:Large language models"
  - "Category:Machine learning"
  - "Category:Prompt engineering"
  - "Category:Technology"
revision_id: 246
wiki_created_at: 2026-09-06T22:19:27Z
wiki_modified_at: 2026-09-06T22:19:27Z
downloaded_at: 2026-09-07T22:22:09Z
---

# Meta Prompting

**Meta-prompting** is an advanced prompt engineering technique in which large language models (LLMs) are used to create, adjust, and optimize their own instructions (prompts)<sup>[\[1\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-cookbook_openai-1)</sup>. Unlike traditional approaches where a human manually writes detailed instructions, meta-prompting focuses on defining the **problem-solving structure** and the roles the model should perform. In other words, it describes *how* to solve a task, not *what* to do in each specific case<sup>[\[2\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-intuition_labs-2)</sup>.

This approach allows LLMs to independently decompose complex problems, clarify queries, and iteratively improve their answers, paving the way for the creation of more autonomous and adaptive artificial intelligence systems.

## Key Methods and Approaches

The term "meta-prompting" gained traction in scientific literature in late 2023 and early 2024, when several research groups introduced similar approaches to managing LLMs.

### The "Scaffolding" Method

One of the key approaches was proposed by researchers from Stanford University and OpenAI<sup>[\[3\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-suzgun2024-3)</sup>. In this framework, called **scaffolding**, the same language model (GPT-4 in the experiments) plays several roles simultaneously:

- **Conductor**: Receives a high-level meta-prompt and breaks down a complex task into a series of simpler subtasks.
- **Experts**: The Conductor initializes several instances of itself as "experts," each of which solves a specific subtask according to specially provided instructions.
- **Integration**: The Conductor coordinates the work of the "experts" and integrates their responses into a final solution.

This orchestration-based approach has significantly improved the efficiency of solving complex tasks. For example, in experiments, meta-prompting outperformed standard single-step queries by **17.1%** and other advanced methods by 15–17% on task sets requiring multi-component reasoning (e.g., the Game of 24, chess puzzles)<sup>[\[3\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-suzgun2024-3)</sup>. Importantly, this method is task-agnostic and operates in a *zero-shot* manner, requiring no specific examples for each new task.

### Theoretical Approach and Structural Patterns

Concurrently, a research group from Tsinghua University proposed its own concept of meta-prompting, shifting the focus from the content of the task to its **syntax and the form** of data representation<sup>[\[4\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-zhang2024-4)</sup>. Building on a categorical model, they demonstrated that an abstract description of a problem's structure allows the model to construct reasoning close in depth to human reasoning and to effectively decompose complex questions into simpler steps.

### Related Concepts

The ideas behind meta-prompting also appear in other methods:

- **[Automatic Prompt Engineer](https://systems-analysis.info/eng/Automatic_Prompt_Engineer_(APE) "Automatic Prompt Engineer (APE)") (APE)**: A method where an LLM automatically generates and selects effective instructions for itself, evaluating them based on the quality of the resulting outputs<sup>[\[5\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-ape2022-5)</sup>.
- **[Self-Refine](https://systems-analysis.info/eng/Self-Refine_Prompting "Self-Refine Prompting")**: An approach where the model iteratively improves its own response by critically analyzing the previous version and generating corrections based on this critique<sup>[\[6\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-self_refine2023-6)</sup>.

## Application and Advantages

Meta-prompting has shown high efficiency in tasks that require complex, multi-component reasoning, such as mathematical proofs, programming, and step-by-step puzzle solving. Key advantages of the method include:

- **Token Efficiency**: Focusing on the general structure of a task instead of listing numerous examples reduces the prompt's length. A meta-prompt acts as a universal template, requiring fewer tokens<sup>[\[7\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-prompting_guide-7)</sup>.
- **Stability and Impartiality**: The method avoids reliance on specific examples, making the model less susceptible to bias from the particulars of the training samples<sup>[\[7\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-prompting_guide-7)</sup>.
- **Dynamic Adaptation**: Unlike a static prompt, meta-prompting allows for iterative improvement. The model can refine instructions, request missing information, and adjust its strategy during the problem-solving process<sup>[\[2\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-intuition_labs-2)</sup>.
- **Generalization to New Tasks**: High-level instructions are more easily transferable to new, unseen tasks, making meta-prompting an advanced form of the *zero-shot* approach.

## Limitations and Risks

- **Computational Cost and Complexity**: The method requires multiple calls to the model for a single task, which increases time and API request costs. Current implementations run sequentially, making parallelization difficult<sup>[\[8\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-decoder_article-8)</sup>.
- **Dependence on Model Capabilities**: The effectiveness of meta-prompting is highly dependent on the quality of the base LLM. Studies have shown that GPT-3.5 gains almost no benefit from this method, whereas for GPT-4 and later models, the advantage is significantly greater<sup>[\[8\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-decoder_article-8)</sup>.
- **Fundamental LLM Limitations**: Meta-prompting is not a panacea. If a task is outside the knowledge base of the underlying model, even a perfectly formulated meta-prompt does not guarantee a correct result. In such cases, either fine-tuning or integration with external tools (such as web search or a code interpreter) is required<sup>[\[3\]](https://systems-analysis.info/eng/Meta_Prompting#cite_note-suzgun2024-3)</sup>.

## External links

- <a href="https://cookbook.openai.com/examples/enhance_your_prompts_with_meta_prompting" class="external text" rel="nofollow">Meta-prompting example in the OpenAI Cookbook</a>
- <a href="https://www.promptingguide.ai/techniques/meta-prompting" class="external text" rel="nofollow">Description of the method in the Prompt Engineering Guide</a>

## Literature

- Suzgun, M.; Kalai, A. T. (2024). *Meta-Prompting: Enhancing Language Models with Task-Agnostic Scaffolding*. <a href="https://arxiv.org/abs/2401.12954" class="external text" rel="nofollow">arXiv:2401.12954</a>.
- Zhang, Y.; Yuan, Y.; Yao, A. C.-C. (2023). *Meta Prompting for AGI Systems*. <a href="https://arxiv.org/abs/2311.11482" class="external text" rel="nofollow">arXiv:2311.11482</a>.
- Zhou, Y. et al. (2022). *Large Language Models Are Human-Level Prompt Engineers*. <a href="https://arxiv.org/abs/2211.01910" class="external text" rel="nofollow">arXiv:2211.01910</a>.
- Madaan, A. et al. (2023). *Self-Refine: Iterative Refinement with Self-Feedback*. <a href="https://arxiv.org/abs/2303.17651" class="external text" rel="nofollow">arXiv:2303.17651</a>.
- Ning, X. et al. (2023). *Skeleton-of-Thought: Prompting LLMs for Efficient Parallel Generation*. <a href="https://arxiv.org/abs/2307.15337" class="external text" rel="nofollow">arXiv:2307.15337</a>.
- Chen, X. et al. (2023). *Universal Self-Consistency for Large Language Model Generation*. <a href="https://arxiv.org/abs/2311.17311" class="external text" rel="nofollow">arXiv:2311.17311</a>.
- Wang, X. et al. (2022). *Self-Consistency Improves Chain-of-Thought Reasoning in Language Models*. <a href="https://arxiv.org/abs/2203.11171" class="external text" rel="nofollow">arXiv:2203.11171</a>.
- Fernando, C. et al. (2023). *PromptBreeder: Self-Referential Self-Improvement via Prompt Evolution*. <a href="https://arxiv.org/abs/2309.16797" class="external text" rel="nofollow">arXiv:2309.16797</a>.
- Zhou, P. et al. (2024). *Self-DISCOVER: Large Language Models Self-Compose Reasoning Structures*. <a href="https://arxiv.org/abs/2402.03620" class="external text" rel="nofollow">arXiv:2402.03620</a>.
- Chen, J. et al. (2024). *Thought-Augmented Reasoning with Large Language Models*. <a href="https://arxiv.org/abs/2406.04271" class="external text" rel="nofollow">arXiv:2406.04271</a>.

## References

1.  <span id="cite_note-cookbook_openai-1">[↑](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-cookbook_openai_1-0) "Enhance your prompts with meta prompting". *OpenAI Cookbook*. <a href="https://cookbook.openai.com/examples/enhance_your_prompts_with_meta_prompting" class="external autonumber" rel="nofollow">[1]</a></span>
2.  <span id="cite_note-intuition_labs-2">↑ <sup>[2.0](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-intuition_labs_2-0)</sup> <sup>[2.1](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-intuition_labs_2-1)</sup> "Meta-Prompting: LLMs Crafting & Enhancing Their Own Prompts". *Intuition Labs*. <a href="https://intuitionlabs.ai/articles/meta-prompting-llm-self-optimization" class="external autonumber" rel="nofollow">[2]</a></span>
3.  <span id="cite_note-suzgun2024-3">↑ <sup>[3.0](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-suzgun2024_3-0)</sup> <sup>[3.1](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-suzgun2024_3-1)</sup> <sup>[3.2](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-suzgun2024_3-2)</sup> Suzgun, Mirac; Kalai, Adam T. (2024). "Meta-Prompting: Enhancing Language Models with Task-Agnostic Scaffolding". *arXiv:2401.12954*.</span>
4.  <span id="cite_note-zhang2024-4">[↑](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-zhang2024_4-0) Zhang, Yifan, et al. (2023). "Meta Prompting for AGI Systems". *arXiv:2311.11482*.</span>
5.  <span id="cite_note-ape2022-5">[↑](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-ape2022_5-0) Zhou, Y., et al. (2022). "Large Language Models Are Human-Level Prompt Engineers". *arXiv:2211.01910*.</span>
6.  <span id="cite_note-self_refine2023-6">[↑](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-self_refine2023_6-0) Madaan, A., et al. (2023). "Self-Refine: Iterative Refinement with Self-Feedback". *arXiv:2303.17651*.</span>
7.  <span id="cite_note-prompting_guide-7">↑ <sup>[7.0](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-prompting_guide_7-0)</sup> <sup>[7.1](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-prompting_guide_7-1)</sup> "Meta Prompting". *Prompt Engineering Guide*. <a href="https://www.promptingguide.ai/techniques/meta-prompting" class="external autonumber" rel="nofollow">[3]</a></span>
8.  <span id="cite_note-decoder_article-8">↑ <sup>[8.0](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-decoder_article_8-0)</sup> <sup>[8.1](https://systems-analysis.info/eng/Meta_Prompting#cite_ref-decoder_article_8-1)</sup> "AI within an AI: Meta-prompting can improve the reasoning capabilities of large language models". *The Decoder*.</span>