›_ Codex Issues
Codex Issues / #9910

Hello. I recently discovered a problem

Resolved 💬 2 comments Opened Jan 26, 2026 by a906511611-ops Closed Jan 26, 2026
💡 Likely answer: A maintainer (etraut-openai, contributor) responded on this thread — see the highlighted reply below.

What version of Codex is running?

OpenAI Codex (v0.89.0)

What subscription do you have?

pro

Which model were you using?

gpt-5.2-codex xhigh

What platform is your computer?

win10

What terminal emulator and version are you using (if applicable)?

Windows Terminal

What issue are you seeing?

I recently discovered a problem. Last week, I used gpt-5.2-codex xhigh to analyze and locate bugs in my game, fix game jitter, improve game fluency, and so on. The analysis by gpt-5.2-codex xhigh is extremely accurate, almost 100% capable of pinpointing the problem and achieving perfect repairs.However, starting from the past few days, I have noticed that the speed of gpt-5.2-codex xhigh has become significantly faster during my use of codex5.2. Typically, it takes more than 10 attempts to locate the problem. On one occasion, I repeatedly tried to use gpt-5.2-codex xhigh to identify a bug that prevented characters from moving in the game, and it took me 10 attempts before finally locating the real bug. This is a serious waste of time. I would like to know if you have made any modifications to the model in the background or if it has become less intelligent.

What steps can reproduce the bug?

I recently discovered a problem. Last week, I used gpt-5.2-codex xhigh to analyze and locate bugs in my game, fix game jitter, improve game fluency, and so on. The analysis by gpt-5.2-codex xhigh is extremely accurate, almost 100% capable of pinpointing the problem and achieving perfect repairs.However, starting from the past few days, I have noticed that the speed of gpt-5.2-codex xhigh has become significantly faster during my use of codex5.2. Typically, it takes more than 10 attempts to locate the problem. On one occasion, I repeatedly tried to use gpt-5.2-codex xhigh to identify a bug that prevented characters from moving in the game, and it took me 10 attempts before finally locating the real bug. This is a serious waste of time. I would like to know if you have made any modifications to the model in the background or if it has become less intelligent.

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

2 Comments

a906511611-ops · 7 months ago

I would like to know how to use the previously slow gpt-5.2-codex xhigh version, as this version was incredibly useful. It was 100% accurate, capable of precise positioning, and could fix code errors!

etraut-openai contributor · 7 months ago

Thanks for the report. No, we haven't made any changes to the model in the background.

There is going to be some natural variation from run to run because the models are stochastic.

In the future, if you see model behavior that you'd like to report, please use the /feedback command to upload your session details. That information is used by the team responsible for training our next-generation models.