How to Use the Right AI Model for Greener Use

Most of us open ChatGPT, Gemini, Claude or Copilot and use whatever setting comes up. More and more often, that means a large model that “thinks” before it answers. Using that to tidy an email is like hiring a removals van to post a letter.

Choosing a smaller, faster setting is one of the simplest Green AI habits there is. It costs nothing, takes a second, and for most everyday jobs the answer is just as good.

Small and large field maple leaves against a dark background
Smaller often does the job just as well.

What are “smaller” and “faster” settings?

Most AI apps let you choose how hard the AI works, usually from a menu next to the message box. There are two main choices:

Why it matters

Every answer an AI gives uses electricity in a data centre. Bigger models and Thinking modes do much more work for each answer, so they use much more energy.

One quick question makes little difference on its own. But habits add up across a day, a team and a year, and choosing the right setting is the easiest habit to change.

How to use smaller, faster AI day to day

Start with the fast setting and step up only when you need to. Here’s how in each app. The names change often, so check yours.

Then:

  1. Step up only if you need to. If the answer isn’t good enough, try again with a bigger model or Thinking on.
  2. Go back to the fast setting for your next everyday task.
  3. Save the big models for hard problems. Complex analysis, maths and multi-step research.

Which setting for which job?

A rough guide, not a rule.
JobSetting
Tidy or rewrite an emailFast, Thinking off
Summarise a documentFast, Thinking off
Brainstorm ideasFast, Thinking off
Hard maths or data analysisThinking on, or a bigger model
Complex researchBigger model with Thinking on

What we don’t know yet

AI companies don’t publish energy figures for each setting, so the numbers above are independent estimates, not exact measurements. Those estimates are improving all the time, and we’ll update this guide as they do.

It’s also worth remembering that easier AI can mean more AI. Choosing a smaller model helps most when we keep asking whether we need AI for the task at all.

Frequently asked questions

Is ChatGPT Instant greener than Thinking?

We don’t know for certain. OpenAI doesn’t publish energy figures for Instant or Thinking, and nobody outside the company has measured them. But related research points the same way: estimates suggest top-tier models use nearly 4 times the energy of fast, lightweight ones, and tests on open AI models found reasoning used 30 times more energy on average. So Instant is likely the greener choice for everyday tasks, but that’s an informed judgement, not a measurement.

Is a smaller model less accurate?

For hard problems, often yes. For everyday jobs like rewriting, summarising and brainstorming, a fast model usually does just as well.

What if my app chooses the model for me?

Settings like Auto in Gemini or Smart in Copilot pick a model for each question. They’re convenient, but you can’t see which model answered. If you want to be sure, choose the fast option yourself.

Getting started

Next time you open your AI app, check the setting before you type. Start small, step up when you need to, and let the habit grow from there.

For more simple ways to get good answers with less waste, try our free Green AI Prompting Guide. If your team is working out how to use AI well, add model choice to your responsible AI policy, or book a free call and we can talk it through.

Green AI