Burned through 6% usage in 90 minutes #39763

What version of the Codex App are you using (From “About Codex” dialog)?

Version 26.814.41407

What subscription do you have?

20x

What platform is your computer?

MacOS

What issue are you seeing?

I am burning through a shocking amount of usage in 90 minutes. 6% of my weekly and I dont feel Ive done particularly taxing tasks. WHen I asked Codex the reply said I had sent up an unusually large amount of data - but it was a single comment of "How can I be down 6% of my daily tokens already? Ive only been working for 90 minutes.".

On further questioning I got:
You’re correct: your prompt was tiny. I described it poorly.
What I received alongside it included app/runtime context such as:

• System and developer instructions
• Your workspace’s AGENTS.md
• Environment and permission details
• The installed-skill and tool catalog
• The recommended-plugin catalog
• Workspace memory guidance
  You did not type that material; Codex attached it automatically. OpenAI’s documentation confirms that prompts, chat history, files, tool results, and AGENTS.md content can all consume tokens. It also notes that cached input is charged at a substantially lower rate. Official usage documentation
  However, I cannot see your account’s detailed usage breakdown, so asserting that this task’s attached context materially caused your 6% consumption was unjustified. It could instead reflect earlier tasks, model/reasoning choice, tool activity, output tokens, or shared ChatGPT Work usage. My earlier explanation presented speculation too confidently.

What steps can reproduce the bug?

Feedback ID: 01a0203f-c3ee-7f50-9226-7dff3303db17

What is the expected behavior?

No response

Additional information

No response