Why is auto review using up so much usage??? I am considering not using it anymore… auto review should honestly just be free.
Can we potentially get an answer from @OpenAI_Support?
If you are on windows, considering looking into this:
I am actually using it on a linux VM but accessing it through Windows. So, I’d assume that means this is not for me
I see. Well, in this case, what I’d do is locate a past thread that was a bit longer and may have triggered the auto-review, and get its thread ID (type /status).
Then, open a new thread with gpt-5.6-luna and ask something like this:
please have a look into thread xxxxx, and analyze if codex-auto-review is being called and why.
This is how I found out what was going wrong in my case and eventually found a solution to fix it. Perhaps it could help you identify what’s happening in your case as well.
If you do, please share it with us. 
One thing to keep in mind is that auto review is just another agent instance with a different prompt - churning over all of your data, second guessing the choices of your main instance - so it’s not surprising it’s using almost as much as your main instance.
I personally have it off, and just give the model restful sandbox tools to do whatever it needs to do.