AI image generation carbon calculator
Generating an image is the most energy hungry thing most people do with AI. A study of 88 models measured an average of 2.9 Wh per image, close to a fifth of a phone charge.
Result
803 g CO2CO2 per year
- Energy per week
- 72.5 Wh
- Energy per year
- 3.77 kWh
- Full phone charges per year
- 251
- Same energy as this many text prompts
- 11,100
These are estimates. Real figures depend on the model, the data center and the length of the answer.
How this is calculated
- Luccioni, Jernite and Strubell (2023), Power Hungry Processing. Image generation averages 2.9 kWh per 1,000 images. arxiv.org
- OpenAI (2025). Sam Altman states an average ChatGPT query uses about 0.34 Wh and 0.32 mL of water. blog.samaltman.com
- Ember, Global Electricity Review 2025. Carbon intensity of electricity in 2024 (Poland 2023). France and Sweden come from a study based on Ember data and are approximate. ember-energy.org
- Our assumptions. A phone charge is 15 Wh, an LED bulb is 10 W, boiling one litre of water takes 110 Wh, a petrol car emits 120 g CO2 per km.
Questions
How much energy does one AI image use?
About 2.9 Wh on average in the Luccioni study, with large differences between models. The most efficient image models used a small fraction of that.
What about AI video?
Video uses far more energy than a still image because the model produces many frames. We do not include it here because no comparable published figure exists yet.