You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Codex 5.3 Disconnecting and Running Extremely Slowly #11271
What version of the Codex App are you using (From “About Codex” dialog)?
MacOS Version 260208.1016 (571)
What subscription do you have?
Pro
What issue are you seeing?
Was running Codex 5.3 on the Mac OS app for the last 3 days without issue. Set the thinking setting to Extremly High, still no problems, but as of around 5:00 PM PST today (Feb 9), all of a sudden, the thread hangs with no updates on the page. I left it like that for an hour and came back and still no updates/messages on the screen so I quite the app and restarted. Prompted it to "retry" and it still hangs on "Thinking", showing messages like "Reconnecting 1/5". Eventually, it did start showing messages, but it's much, much slower than before. I have a skill to just clean up the local git state and switch back to main. This was consistently taking about 10s or less. Now It takes almost 8 minutes. Any ideas what happened tonight?
What steps can reproduce the bug?
Running the Codex app on Mac.
What is the expected behavior?
It should be thinking and working 10x faster than it is.
Thanks for the bug report. If you're able to repro this, please use the /feedback slash command to upload your logs and session details, then post the thread ID here. I can take a look at the data and see if I can diagnose.
Since the release of the Codex 5.3 API, this issue has started to occur.
At the very least, I hope that users of the Codex CLI and app are not experiencing slower performance.
What’s most concerning is that the performance is actually worse than Codex 5.2.
@jskorlol, if you're seeing slow performance and think something is wrong, please use /feedback to upload your logs and session details and post your thread ID.
Adding some screenshots of what I'm seeing. A simple prompt to merge a PR and clean the local git state was taking about 10 seconds before last night, but now is in the order of several minutes. When asked to investigate why it's taking that long, Codex claims it's not taking that long and that there is either a lot of latency in the chat or network delays. I'm on gig speed internet, in a metro area and it's been running fine. I would like the app to record the chat latency so we can get a better sense of where the time is being spent.
It's back to normal now. The other thread that @MisterRound linked (#11189) basically identifies the culprit which is OpenAI flagging our accounts as potentially be using the models for cyber attacks so we have to submit government issued IDs to gain back access. Not sure why I got flagged. I was using it heavily for a few days to work on my B2C SaaS.
The applicatioin is unusable on my m2 ultra 64gb. Switching to different threads takes ages. it runs very buggy. the AI part is super quick however the interface is close to useless. Losing my nerf here. when you submit a question you cannot be sure it was actually submitted.
I'm also seeing incredibly slow performance. It's painful. Clicking into a thread has a noticeable lag, and everything is just grinding to a halt. I have a high-spec MacBook Pro with plenty of RAM; it should be fine but performance in the Mac app just seems to be really really poor after just 10 mins of usage.
Reacted by gulla0, Felipe Hertzer, Paweł Dębiński, Zhanarys and solo812
A bunch of different issues were discussed in this thread. I think we've addressed most (all?) of them. I'm going to close this bug report. If you're still seeing any of these issues, please open a new bug report so we can track it separately.
Activity
etraut-openai commentedon Feb 10, 2026
Thanks for the bug report. If you're able to repro this, please use the
/feedbackslash command to upload your logs and session details, then post the thread ID here. I can take a look at the data and see if I can diagnose.jskorlol commentedon Feb 10, 2026
Since the release of the Codex 5.3 API, this issue has started to occur.
At the very least, I hope that users of the Codex CLI and app are not experiencing slower performance.
What’s most concerning is that the performance is actually worse than Codex 5.2.
etraut-openai commentedon Feb 10, 2026
@jskorlol, if you're seeing slow performance and think something is wrong, please use
/feedbackto upload your logs and session details and post your thread ID.jskorlol commentedon Feb 10, 2026
@etraut-openai
019c45f6-20eb-7251-bc16-20f18f4a76fe
019c45bb-f366-7b72-9d00-1dce6f0d7df5
It’s become slower and the performance is worse than it was in 5.2.
jonlave commentedon Feb 10, 2026
@etraut-openai - here's my feedback id's. Thanks
019c452d-5845-7491-ab3b-cb33cc71610a
019c451a-ff3e-7cf0-85ae-a2ab496d8228
pbowyer commentedon Feb 10, 2026
And mine: 019c4758-d74a-7df3-bda4-524283b1e298
christian-wiloejo commentedon Feb 10, 2026
@etraut-openai Here's my feedback ID: 019bfb21-b535-71a3-94fa-ee279c55a1b5
marjan2k commentedon Feb 10, 2026
@etraut-openai some of my feedback IDs:
IlyaShorin commentedon Feb 10, 2026
same here 019c47ce-36ff-7e21-a537-ac6544097883
jonlave commentedon Feb 10, 2026
Adding some screenshots of what I'm seeing. A simple prompt to merge a PR and clean the local git state was taking about 10 seconds before last night, but now is in the order of several minutes. When asked to investigate why it's taking that long, Codex claims it's not taking that long and that there is either a lot of latency in the chat or network delays. I'm on gig speed internet, in a metro area and it's been running fine. I would like the app to record the chat latency so we can get a better sense of where the time is being spent.
MisterRound commentedon Feb 10, 2026
Likely related to #11189 assume you're all actually seeing redirected 5.2 responses
jonlave commentedon Feb 10, 2026
It's back to normal now. The other thread that @MisterRound linked (#11189) basically identifies the culprit which is OpenAI flagging our accounts as potentially be using the models for cyber attacks so we have to submit government issued IDs to gain back access. Not sure why I got flagged. I was using it heavily for a few days to work on my B2C SaaS.
Karimrekiki commentedon Feb 11, 2026
The applicatioin is unusable on my m2 ultra 64gb. Switching to different threads takes ages. it runs very buggy. the AI part is super quick however the interface is close to useless. Losing my nerf here. when you submit a question you cannot be sure it was actually submitted.
benjamindell commentedon Feb 11, 2026
I'm also seeing incredibly slow performance. It's painful. Clicking into a thread has a noticeable lag, and everything is just grinding to a halt. I have a high-spec MacBook Pro with plenty of RAM; it should be fine but performance in the Mac app just seems to be really really poor after just 10 mins of usage.
JamesIV4 commentedon Feb 13, 2026
I'm having this issue today. It's taking many minutes to make very simple changes. 019c5941-9dce-7b11-8eb8-ee6d41759d8e
etraut-openai commentedon Feb 20, 2026
A bunch of different issues were discussed in this thread. I think we've addressed most (all?) of them. I'm going to close this bug report. If you're still seeing any of these issues, please open a new bug report so we can track it separately.