My AI footprint
The carbon footprint of a year of using AI, calculated task by task, with the effect of your settings.
Describe your use
assistants, week, tasks you give to AI
Calculate your footprint
annual footprint (CO2e), range, equivalents
Find your settings
the effect of each lever
Sourced factors from the AI Database by Projet Celsius
Question 1 of 4 · 3 minutes
Which assistants do you use?
Every one you open at least once a month.
No account
Nothing is sent
Sourced factors
What My AI footprint calculates, from your request to your year
A calculation of the electricity used by the servers and of its life-cycle emissions, in 6 steps, each with its sourced factor in the AI Database.
You send a request
a question, a document to summarise, an image to create
The model reads and writes tokens
this is inference; a token is a piece of a word, and half a page of answer contains about 400
The server's chips compute
the larger the model, the more it uses; extended reasoning multiplies it by 10 in the AI Database (× 1.9 to × 30 on average in published measurements)
The data centre cools down
cooling and building losses, at the level of the large operators (× 1.10)
Electricity emits greenhouse gases
depending on the country of the servers, from France to the world mix
Your year
the sum of all your requests, task by task
Your useComputing on the serversElectricity and climate
Not included: model training, server manufacture, the network and your device. Their orders of magnitude are in the AI Database. Record of the quick question (in French)
A year with AI, calculated live
The hub's estimator, and 6 figures that count in your year.
Estimate my year
Tool
Use37%
What weighs most: deep research
270 g over the year
Electricity× 11
The country of the servers
from 41 g per kWh in France to 458 worldwide
AI Database0.14 g
A quick question
0.3 Wh, medium model, no reasoning
Settings× 10
Extended reasoning mode
electricity for the same question, in the AI Database
Benchmark5.2 km
By petrol or diesel car
that is 85 phone charges
Lever- 140 g
The typical user's first lever
Deep research for real investigations
The footprint of your AI, estimated in 3 minutes
From your tools to your settings, in 4 steps
Every answer you do not know keeps the typical user's value. Your year builds up as you answer.
My tools
My weekWhat I doMy settingsMy results
My tools
The assistants you open every month and their plan; the country of their servers enters the calculation.
My week with AI
The days you use it and the number of requests; the year follows, holidays included.
What I do
Writing, searching, reading a document, coding, creating an image: each task has its frequency and its weight.
My settings
Reasoning mode and model size, shown in electricity before you choose.
Calculate the footprint of my year with AI
3 minutes, no account, nothing is sent. At the end: your total, what makes it up and the settings that reduce it.
Going further, in the hub
Cropped photos, Wikimedia Commons : Tran Mau Tri Tam tranmautritam ↗ (CC0 ↗) ; Shixart1985 ↗ (CC BY 2.0 ↗) ; John Robert McPherson ↗ (CC BY-SA 4.0 ↗) ; Kallie Hagel ↗ (CC BY-SA 4.0 ↗). Photos under CC BY-SA remain under CC BY-SA 4.0 once cropped.
.jpg){kind=link}
