What the report measures
OpenAI’s October 14 report says Codex and ChatGPT Work accounted for more than half of output tokens in its EU ChatGPT Enterprise and Business account sample by the end of September. Tokens are pieces of text processed or generated by a model.
The report also shows 8.1 times as many weekly active agent users among EU small and medium businesses compared with early May. It defines those businesses as having 200 or fewer associated users. These are OpenAI’s measurements, not a survey of every European business.
What the numbers do not show
More generated text does not, by itself, establish time saved, accuracy, or profit. The report’s usage measures should not be read as proof that an individual company will benefit by the same amount.
Our take for general readers
Editorial analysis: start with a task you can check, such as turning fictional notes into a draft checklist. Compare the time spent reviewing and correcting the output with doing it yourself. A useful result matters more than producing more text or granting an agent more access.
Source timing
This report was newly observed in our afternoon check. Its October 14 date does not establish whether it appeared after our previous source check.