What does your team's AI use weigh?

We run CrbonFree on our own accounts. Over the past year our team used 33.4 billion tokens, which came to 2.64 tonnes of CO2e and 858 kWh of electricity. The dashboard compares that to driving a typical sedan for a year. Almost all of it was input: 33.3 billion tokens read against 94 million written.

We're curious where other teams land. If you've tried to put a number on your AI use, what did you get, and what made it hard? And what would you need to see before trusting a figure like this in a sustainability report?

Every factor we use is on our methodology page, so push back on anything that looks off.