Codex usage got decreased by 2-4 times.

[richard-mayers]

What issue are you seeing?

The current token burning rate increased a lot in the last 2-3 days.

Reddit is full of posts about this. So this is not just perception.

I also track my token usage and the % variation daily. And it clearly has changed without OpenAI notifying anything.

What happened?

What steps can reproduce the bug?

Track your tokens and % daily and you can see the burning rate changes.

What is the expected behavior?

No response

Additional information

No response

[github-actions]

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

• Rate Limits Draining Faster #12728
• Double usage limits removed on my account? #12759
• Codex usage limit without using it #12799
• Usage limits warning message appears despite limits not being close to exhaused #12851
• "You've hit your usage limit." Despite 10% Rate Limit Usage Remaining #12299

Powered by Codex Action

[abulgher]

We all had the same impression. They claimed they have not changed anything.

[etraut-openai]

We have not changed usage limits. If you are experiencing higher burn rates, then you are using more tokens. We have been rolling out various performance improvements, so you should see more work done by your agent in a shorter amount of time. That may contribute to the impression that usage is disappearing faster than before.

[smileBeda]

How’s it possible that an avalanche of users suddenly do „more“ than before?
I for once run a benchmark like this:

• same prompt as last werk
• same code base as last week
• 20% of the PRO allowance burned in 55 minutes vs same task last week on PLUS - and plus lasted me a full week.

So, there’s something wrong here

[o6uoq]

Could this be related to version 0.106.0 and using multi-agents ?

[Joe-le-Tekkie]

We are absolutely not making this up. I've been using Codex since inception. My tokens in the last few days are getting burnt unbelievably fast. Usage is no different than before. Have you not considered that something else might be at play here rather than changed usage limits being responsible for this?

[etraut-openai]

We're investigating the issue and have some theories that we're tracking down. We're tracking the issue in #13186.