Estimate your AI footprint
A few questions, an estimate with its range shown honestly, and no account needed. Already running through an API? Connect it and skip the guessing: metered usage produces a much narrower range. Curious where these numbers come from? See the compute model.
Claude Pro and Max, ChatGPT Plus and Pro, and the Gemini app publish no usage API at any permission level, so a model is the only honest instrument. It anchors on the questions you can actually answer.
Count your own messages, not the replies. A rough number is fine; the estimate carries the uncertainty either way.
If you are unsure, leave it on the mix; that routes to the more expensive coefficient rather than the cheaper one.
About 435 messages a month, which we model as 3.9M tokens (range 1.0M to 14.6M).
Water, per month
14.0 gal
1.17 gal – 167 gal
Carbon, per month
7.87 kg CO2e
0.768 kg CO2e – 80.6 kg CO2e
Energy, per month
15.5 kWh
1.58 kWh – 151 kWh
Drawn the way the console draws them: the bar rises to the 90th percentile, which is the accrual basis retirements are sized against; the shade changes at the median; the whisker spans the 90 percent range. Every constant behind these bars is public, with its source: the methodology.
These are self-estimated, not metered. The range is wider than a connected account's because it carries two unknowns: how many tokens your usage actually involves, and what a token costs environmentally. Anything you publish from this figure has to carry the range and say it was self-estimated.
At the central estimate, that is 4M billable tokens a month, which lands in the Free tier.
Free
to 10M
Personal Light
to 5M
Personal
to 25M
Personal Plus
to 60M
Personal Max
to 100M
Calculating costs nothing, at any scale. A free account meters and estimates up to 10M billable tokens a month today; paid tiers, which add automatic monthly certificate retirements sized at the accrual basis, are not yet on sale. Details on pricing.