For the past several days, I had been repeatedly experiencing the following error in Codex:

“Selected model is at capacity. Please try a different model.”

The issue was severe enough that Codex was often practically unusable for me. Tasks would frequently get interrupted after only a few messages, and I also encountered similar capacity-related errors during remote compact tasks.

A short while ago, I tried Codex again and noticed that the interruption/capacity issue seems to have improved. I was finally able to run a task continuously for more than 30 minutes without being interrupted.

However, I immediately noticed a different and potentially serious issue: my Pro 20x usage is now being consumed extremely quickly.

I was using GPT-5.6 Terra Max, and during roughly 30 minutes of continuous use, my Pro 20x usage decreased by approximately 34%.

This level of usage consumption seems far higher than what I normally experience.

I have paused my Codex tasks for now because I don’t want to continue consuming my weekly allowance at this rate.

Has anyone else noticed unusually fast usage depletion after the recent capacity issues improved?