greenai.guide

Home › How do I…? › How do I write prompts that use less energy?

How do I…?

How do I write prompts that use less energy?

Updated · 1 min read

In short

Write short, specific prompts that say exactly what you want, give only the context the model needs, and set a length for the answer. Energy rises with the amount of text a model reads and writes, so trimming both, and getting a usable answer first time, saves the most.

Steps

  1. State the task, the audience and the format in the first line.
  2. Set a length limit, such as "in three bullet points" or "under 100 words".
  3. Paste only the paragraphs or data the task needs, not whole documents.
  4. Give one example of what good looks like if the task is unusual.
  5. Correct a near-miss with a short follow-up instead of starting again.

A good prompt saves energy by being short, specific and right first time. Language models process text as tokens, chunks of roughly part of a word, and every token read or written costs computation. Delavande, Pierrard and Luccioni found that energy grows linearly with prompt and generation length.

What does an efficient prompt look like?

It says what, for whom, and how long. Compare:

  • Wasteful: pasting a 20-page report and asking "thoughts?"
  • Efficient: pasting the two relevant pages and asking "List the three biggest risks in this section for a non-specialist board, one sentence each."

The second reads less, writes less and is more likely to be usable straight away.

How much can it save?

UNESCO and University College London found that shorter prompts and responses could cut energy use by more than half. Input length matters a lot: Epoch AI estimates that a typical query of about 0.3 Wh rises to about 2.5 Wh with a 10,000-token input.

Common mistakes

  • Pasting everything "just in case". Extract the part that matters.
  • Leaving length open. Models tend to write more than you need unless told otherwise.
  • Vague first attempts followed by several retries. One clear prompt usually beats three vague ones.
  • Worrying about "please" and "thank you". Short polite messages do cost a little energy, but long inputs and repeated regeneration matter far more.

If you build tools, you can enforce some of this in code: see how to limit how much an LLM writes. For more everyday tips, read use AI efficiently.

Sources

  1. Smarter, smaller, stronger: resource-efficient generative AI and the future of digital transformation, UNESCO and University College London (2025)
  2. How much energy does ChatGPT use?, Epoch AI (2025)
  3. Small talk, big impact: the energy cost of thanking AI, Delavande, Pierrard and Luccioni (arXiv 2601.22357) (2026)
  4. What are tokens and how to count them?, OpenAI Help Center (2025)

Join us

Help cut AI's impact on the planet

Get new guides, funding calls and directory listings by email, about once a month. Or add your product, course, PhD or research to the directories.

No tracking. Unsubscribe in one click. Other ways to join · Privacy