Codex randomly Freezes in 0.91.0

Resolved 💬 13 comments Opened Jan 26, 2026 by Akrelion45 Closed Jan 27, 2026

💡 Likely answer: A maintainer (etraut-openai, contributor) responded on this thread — see the highlighted reply below.

What version of Codex is running?

0.91.0

What subscription do you have?

Pro

Which model were you using?

GPT-5.2-codex-xhigh

What platform is your computer?

Microsoft Windows NT 10.0.26100.0 x64

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

_No response_

What issue are you seeing?

In the newest version codex randomly freezes after giving it a task. Had it like 3 or 4 times in the last hour. I see there is a backgroun terminal running all the time. Maybe it is related to that.
I did the /feedback command from a 2nd cmd, sadly can't do it in the main one because it froze. But maybe still helpful.
From what i noticed codex is still working, it gave me an output a few minutes later (It did create a file as requested from me in the task) but the cmd is completly unresponsive.

Following Feature flags from my config.toml are enabled:

[features]
unified_exec = true
shell_snapshot = true
collab = true
steer = true
powershell_utf8 = true
collaboration_modes = true

What steps can reproduce the bug?

Uploaded thread: 019bfb47-461f-7522

Newest Codex version.
Probably enable my feature set from the config.toml
Then give it a task on windows

What is the expected behavior?

CMD should not freeze like in previous codex versions

Additional information

_No response_

View original on GitHub ↗

13 Comments

etraut-openai contributor · 5 months ago

You have many experimental features enabled. You must like living life on the bleeding edge!

It's possible that one of these features is related to the hang you're seeing. Since you have a repro, it would be helpful if you could do some bisecting (i.e. disable half of these experimental features) to try to narrow down the cause.

Akrelion45 contributor · 5 months ago

Always willing to test what the codex team is cooking. 😊

I have just disabled collab and collaboration_modes and now its working fine, but of course i need some more time to verify if it was really one of these 2 settings. Could also just be a coincidence now. I will test now for an extended period of time with both of these disabled.
I will let you know in a further response if one of these was / is the problem.

etraut-openai contributor · 5 months ago

OK, thanks. Those close are really helpful!

Akrelion45 contributor · 5 months ago

Alright, it just happened again with collab and collaboration_modes set to false. Seems like they weren't the problem.

But again, codex spawned a background terminal, so it must be related to that setting (I have Background terminal enabled in Experimental features).
When it worked fine over the last hour, i didn't noticed a background terminal was used.

euclideannodes · 5 months ago

same issue

leiDnedyA · 5 months ago

Can confirm this is happening on linux as well. Here's my config.toml, I have 0 experimental features enabled (I've redacted some local file paths)

model = "gpt-5.2-codex"
[projects."REDACTED"]
trust_level = "trusted"

[projects."REDACTED"]
trust_level = "trusted"

[notice.model_migrations]
gpt-5-codex = "gpt-5.2-codex"

Here's my uname -a output as well in case that's useful

Linux aydens-laptop 6.8.0-90-generic #91~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Nov 20 15:20:45 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

Lmk if there's anything else you need from me 🫡

thoppe · 5 months ago

Confirming it's happening to me on Linux as well, here's my redacted config.toml. It seems to happen in one out of every 3 prompts.

model = "gpt-5.2-codex"
model_reasoning_effort = "medium"

[projects."/home/travis/.................."]
trust_level = "trusted"

[notice]
"hide_gpt-5.1-codex-max_migration_prompt" = true

[notice.model_migrations]
"gpt-5.1-codex-max" = "gpt-5.2-codex"

╭────────────────────────────────────────────────────╮
│ >_ OpenAI Codex (v0.91.0)                          │
│                                                    │
│ model:     gpt-5.2-codex medium   /model to change │
│ directory: ~/.codex                                │
╰────────────────────────────────────────────────────╯

etraut-openai contributor · 5 months ago

Thanks to everyone who reported this issue. We think we understand the root cause and have a fix available in prerelease version 0.92.0-alpha.7. We'd appreciate it if you'd give it a try and let us know if it appears to fix the issue for you.

patricksuo · 5 months ago

Confirmed on my side 👍

I had a local repro where Codex froze consistently (3/3 runs) on v0.91.0.
After upgrading to the latest prerelease, I can no longer reproduce the freeze so far.

jif-oai contributor · 5 months ago

I just released 0.92.0 that should contain the fix, let me know if you still get some freezes

PiquoInterest · 5 months ago

going to check @jif-oai and will let you know! Have updated to 0.92.0, it has been fine thus far.

etraut-openai contributor · 5 months ago

We've received multiple internal and external confirmations that the hang is fixed in the latest release, so I'm going to close the issue. Thanks to everyone who reported the issue and helped us narrow it down!

V391 · 2 months ago

its may 7th, 2026 and I am still having this same issue.