In short
Count roughly how many text questions, long documents and images you use each week, multiply by published per-request figures, and you have a rough range. A typical text prompt is a fraction of a watt-hour on the companies' own figures, while long inputs and images cost far more, so your heaviest habits usually dominate the total. Treat the result as an estimate, not a precise number.
Steps
- Note for a typical week how many text questions, long-document tasks and images you ask for.
- Multiply each by a published per-request figure, using a low and a high value.
- Add the results to get a weekly energy range in watt-hours.
- Convert to carbon with your local grid's carbon intensity if you want a carbon figure.
- Look at which habit dominates and change that one first.
A back-of-envelope estimate is enough to show which of your habits matter. You can't meter a chatbot, but published figures let you build a sensible range.
Which figures can I use?
| Type of use | Published figure | Source |
|---|---|---|
| Typical text question | about 0.24–0.34 Wh | Google (median Gemini text prompt); OpenAI (average ChatGPT query) |
| Text question, independent estimate | about 0.3 Wh | Epoch AI |
| Very long input (about 100,000 tokens, roughly a book) | almost 40 Wh | Epoch AI |
| Generating an image | about 2.9 kWh per 1,000 images on average, so about 2.9 Wh each | Luccioni, Jernite and Strubell |
The company figures are self-reported and use different boundaries; the image figure comes from a 2024 study of open models. Use them as rough guides.
A worked example
Fifty short questions a week at 0.3 Wh is about 15 Wh. Add ten image generations at about 2.9 Wh each and you add roughly 29 Wh: the images more than double the total. Pasting a few very long documents could add more again. To convert to carbon, multiply by the carbon intensity of your electricity grid; see finding grid carbon intensity, though the data centre's grid matters more than yours.
What makes the biggest difference
Images, video, very long inputs and endless regeneration, not the number of short questions. Saying "please" barely registers. For scale, see putting your AI energy use in perspective, and for habits that help, reducing the energy of your ChatGPT use.
The green AI guide for everyone explains why the figures vary.
Sources
- Measuring the environmental impact of delivering AI at Google scale, Elsworth et al., Google (2025)
- The gentle singularity, Sam Altman (OpenAI) (2025)
- How much energy does ChatGPT use?, Epoch AI (2025)
- Power Hungry Processing: Watts Driving the Cost of AI Deployment?, Luccioni, Jernite and Strubell, ACM FAccT (2024)
Related
How do I put my AI energy use in perspective?
How one AI question compares with everyday energy use, why images, video and long documents change the picture, and why the global total still matters.
How do I reduce the energy my ChatGPT use consumes?
Practical ways to cut the energy of using ChatGPT and similar chatbots: shorter exchanges, fewer regenerations, lighter models and less image generation.
How do I check a claim about AI's energy use?
Six questions to test any claim about AI energy, water or carbon: per what, who says so, what's included, how old, compared with what, total or per use.
Green AI: a plain-English guide for everyone
What using AI really costs in energy and water, which everyday habits matter and which don't, how to judge claims, and how to have your say.