TokenSave Data

Measurements we ran ourselves on AI tokens and costs, published as open datasets. Each page has the headline numbers, the full table, a CSV download, the method and its limits, and a ready-made citation. Everything is free to reuse under CC BY 4.0 with a link back.

• Token cost by language: 41 languages measured on GPT tokenizers The same prompt in 41 languages: +3% to +144% more GPT tokens than English. CSV · updated 2026-10-16

More datasets are on the way. Using our data in an article or paper? We'd love to hear about it: contact@jonhisking.com