Token drain issue w/ multiple agents

Open 💬 3 comments Opened Jul 1, 2026 by wantseeplanbe

💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

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

Version 26.616.71553 • Released Jun 22, 2026

What subscription do you have?

Plus

What platform is your computer?

MacOS: Darwin 25.5.0 arm64 arm

What issue are you seeing?

My 5-hr compute quota drains from 100% to 0% within 1-2 minutes after codex spins up about 4-5 sub-agents. It also drains my weekly compute by about 20-25% within that same 1-2 min timeframe. Historically I haven't seen any token drain like this, despite using about the same amount of compute as previously.

What steps can reproduce the bug?

Feedback ID: 019dae12-0b40-7800-bfad-82c49ed2a989

What is the expected behavior?

Token drain doesn't occur as rapidly (1-2 mins seems extreme) when Codex uses sub-agents.

Additional information

_No response_

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 1 month ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

#30756
#30407
#29940
#30427
#30246

Powered by Codex Action

mohammedtaha-ai · 1 month ago

I’m seeing a very similar issue in Codex App.

App version: 26.623.13972.0
Plan: Plus
OS: Windows x64
Model/effort: GPT-5.5 High
Scenario: relatively small dependency/import replacement task with 3 subagents
Observed: the 5-hour usage meter dropped from around 90% remaining to around 10%, then shortly after showed that I was out of Codex messages.

This feels disproportionate compared to the task size. It is also unclear whether the main session continues consuming usage while waiting for subagents to finish.

It would help if Codex App showed per-request and per-subagent usage accounting.

eah3699 · 1 month ago

Codex sub-agents are pretty much unusable with these drain rates.