What issue are you seeing?
Summary
Since yesterday, I've noticed a sudden and significant increase in Codex usage consumption on my ChatGPT Plus subscription.
The 5-hour usage quota now depletes considerably faster than it did previously, even though my coding workflow, typical tasks, and model preferences have remained unchanged.
I'm unsure whether this is caused by a change in usage accounting, reduced effective limits, increased token consumption or a regression.
Environment
- Subscription: ChatGPT Plus
- Models primarily used:
- GPT-5.6 Sol (High)
- GPT-6.0 Sol (Medium/High)
- GPT-6.0 Luna (Medium/High)
- Use case: Software development
- First noticed: October 7, 2026
- Affected limit: 5-hour Codex usage window
Actual Behavior
- The 5-hour usage allowance decreases noticeably faster than before. Not sure if also overall token got decreased?
- Comparable coding tasks appear to consume a substantially larger portion of the available quota.
- I reach usage limits much earlier during normal development sessions.
- This change appeared quite suddenly rather than gradually.
- I have not intentionally made major changes to my workflow that would explain the increased consumption.
I do not currently have precise before-and-after token measurements, but the difference in effective usage capacity is noticeable.
Questions
- Were changes introduced to Codex usage limits, token accounting, or model-specific consumption rates around October 6–7, 2026?
- Have the effective 5-hour limits for ChatGPT Plus users changed without an accompanying announcement?
- Anybody else has the same issues?
What steps can reproduce the bug?
Use Codex with ChatGPT Plus subscription.
Select GPT-5.6 Sol (High) or GPT-6.0 Sol/Luna (Medium/High).
Perform typical coding tasks in an existing software project.
Monitor the remaining 5-hour usage allowance.
Compare the rate of quota depletion with usage before October 7, 2026.
What is the expected behavior?
Under the same subscription, with comparable tasks, models, and reasoning settings, usage consumption should remain reasonably consistent unless changes have been introduced to token accounting or usage limits.
Additional information
Just saw this related report:
#51604
Cheers!
What issue are you seeing?
Summary
Since yesterday, I've noticed a sudden and significant increase in Codex usage consumption on my ChatGPT Plus subscription.
The 5-hour usage quota now depletes considerably faster than it did previously, even though my coding workflow, typical tasks, and model preferences have remained unchanged.
I'm unsure whether this is caused by a change in usage accounting, reduced effective limits, increased token consumption or a regression.
Environment
Actual Behavior
I do not currently have precise before-and-after token measurements, but the difference in effective usage capacity is noticeable.
Questions
What steps can reproduce the bug?
Use Codex with ChatGPT Plus subscription.
Select GPT-5.6 Sol (High) or GPT-6.0 Sol/Luna (Medium/High).
Perform typical coding tasks in an existing software project.
Monitor the remaining 5-hour usage allowance.
Compare the rate of quota depletion with usage before October 7, 2026.
What is the expected behavior?
Under the same subscription, with comparable tasks, models, and reasoning settings, usage consumption should remain reasonably consistent unless changes have been introduced to token accounting or usage limits.
Additional information
Just saw this related report:
#51604
Cheers!