What it is
The Prompt Engineering Guide from DAIR.AI collects explanations, examples, papers and notebooks on prompting. Its techniques section runs to 18 pages, from zero-shot and few-shot prompting to chain-of-thought, ReAct, Reflexion and graph prompting, and it has grown sections on AI agents, context engineering, RAG, model notes and risks such as adversarial prompting.
It is published as the website promptingguide.ai, with translations into 13 languages besides English, though the agents section exists only in English.
The repository has no releases. The last commit, on 11 March 2026, pointed the site's announcement bar at DAIR.AI's paid Academy.
Our review
The guide remains the most-linked free reference on prompting, but it has stopped moving. It has about 79,500 GitHub stars and roughly 290 open issues and pull requests, and nothing has been committed since 11 March 2026. That last commit changed the site's banner to promote DAIR.AI Academy, which says something about where the maintainers' attention has gone.
The core is still worth reading. Each technique page explains the idea with a worked example and links the original paper, which makes it a quick way to understand terms like self-consistency or tree of thoughts when they turn up in a design review. The papers, datasets and notebooks sections add a reading list that would take hours to assemble.
Age is the problem. Several pages teach workarounds, such as asking a model to reason step by step, that current reasoning models do unprompted, and the guide rarely says which techniques still pay off on which models. Pages carry no dates, so a reader cannot tell a 2023 page from a 2026 one. The newer agent and context-engineering pages are the freshest material, and they are English only.
The commercial layer is easy to ignore. The site itself is free; DAIR.AI Academy sells a Pro plan at $49 a month or $350 a year, with team pricing for five or more seats on request. Model vendors' own prompting guides are narrower but are revised with each model release, which is the one property this guide has lost.
Where it earns its place
A backend engineer joining a team that ships LLM features
Learn the vocabulary of prompting and RAG before the first design review
Read the techniques and RAG pages in an afternoon, following paper links where a method is in use. It costs nothing; check any advice against the model vendor's current guide, because the pages are undated.
A learning and development lead at a 60-person consultancy
Give consultants a common grounding in prompting without buying a course
Use the free site as the reading list and add the vendor guide for whichever model the firm licenses. If structured courses are wanted, DAIR.AI Academy Pro is $350 a year per person, so ten people cost about $3,500 at list; team pricing is not published.
A PhD student writing a related-work section on agent prompting
Trace techniques like ReAct and Reflexion back to their sources
Technique pages and the papers section link the primary papers, which saves time assembling citations; newer work since early 2026 will be missing.
Before you commit
- Setup
- Nothing to install: read promptingguide.ai or clone the repository. Notebooks need Python and an API key for the model they call.
- Data and privacy
- Reading the guide sends nothing anywhere beyond normal website visits; notebooks send prompts to whichever model provider you configure. Content is MIT-licensed, so it can be reused commercially with attribution.
- Cost at scale
- The guide is free. DAIR.AI Academy, which the site promotes, has a free tier, Pro at $49 a month or $350 a year, and Teams for five or more seats at an unpublished price.
- Limits
- No commits since 11 March 2026, about 290 open issues and pull requests, no dates on pages, and the agents section untranslated.
What's good
- 18 technique pages, each linked to its original paper
- Free, with translations into 13 languages
- Useful papers, datasets and notebooks lists
What's not
- No commits since March 2026
- Pages are undated and some advice predates reasoning models
- Agents section is English only
- Site promotes paid Academy courses
Best for
Anyone who wants a free reference on prompting and agent techniques.
Alternatives
Prompt Engineering Guide: common questions
- Is Prompt Engineering Guide worth it?
- A good free map of where prompting techniques came from, and an unreliable guide to which ones still matter on current models. We rate it 4.0 out of 5.
- How much does Prompt Engineering Guide cost?
- Free and open source under the MIT license. DAIR.AI also sells paid courses separately.
- What are the downsides of Prompt Engineering Guide?
- No commits since March 2026; pages are undated and some advice predates reasoning models; agents section is English only; site promotes paid Academy courses.
- Who is Prompt Engineering Guide best for?
- Anyone who wants a free reference on prompting and agent techniques.
- How much does Prompt Engineering Guide cost for a team or in production?
- The guide is free. DAIR.AI Academy, which the site promotes, has a free tier, Pro at $49 a month or $350 a year, and Teams for five or more seats at an unpublished price.
- Is Prompt Engineering Guide safe for company data?
- Reading the guide sends nothing anywhere beyond normal website visits; notebooks send prompts to whichever model provider you configure. Content is MIT-licensed, so it can be reused commercially with attribution.
- What are the best Prompt Engineering Guide alternatives?
- The closest alternatives we have reviewed are LLM Course and Generative AI for Beginners.
Answers checked Oct 23, 2026.
Make Prompt Engineering Guide? Show this rating on your site
