codex cli taking a very long time to respond to even simple requests

Open 💬 5 comments Opened Mar 15, 2026 by nkdhruw

💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of Codex CLI is running?

v0.114.0

What subscription do you have?

Pro

Which model were you using?

gpt-5.4 (reasoning high, summaries auto)

What platform is your computer?

Linux 5.15.0-168-generic x86_64 x86_64

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

terminal that comes with the jupyterlab

What issue are you seeing?

The responses from the codex have been super slow lately. It is stuck doing nothing. I had to kill my last 3-4 requests as it took over 10 mins with no response from the codex. It flashes the stream disconnected message on the terminal and then again switches to working flash message.

FYI - I had about ~80% context window left in all the cases.

What steps can reproduce the bug?

Not sure if this is reproducible.

What is the expected behavior?

_No response_

Additional information

_No response_

View original on GitHub ↗

5 Comments

github-actions[bot] contributor · 5 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

#14755
#14670
#14629

Powered by Codex Action

nkdhruw · 5 months ago

Uploaded thread: 019cf284-03db-71c2-95cb-a7acd1293b56

Reconnecting... 4/5 (25m 25s • esc to interrupt)
└ Stream disconnected before completion: error sending request for

Sapientropic · 5 months ago

I filed a more tightly scoped follow-up for the still-reproducible March 16, 2026 residual / recurrence where the sharp boundary is new-thread first turn vs existing-thread resume: #14795. The added evidence there includes Windows Desktop app-server logs showing slow model/list, failed to refresh available models: timeout waiting for child process to exit, and fast thread/resume on existing threads. Posting here in case it helps triage overlap.

etraut-openai contributor · 5 months ago

If you're able to repro this, please use /feedback to upload your logs and session details, then post the thread ID here.

arnaudblois · 5 months ago

I’m seeing the same issue today.

Environment

Pop!_OS Linux
Codex CLI + VS Code extension
Latest versions
Example thread ID: 019cf6f5-804f-7250-abf0-401210b28c50

Symptoms

Codex is extremely slow to the point of being mostly unusable.
Even trivial tasks take several minutes before the first action.
Once local tools run (rg, sed, etc.), they execute instantly.
The delay appears to happen before tool execution.

Repro characteristics

Happens with multiple models: gpt-5.4, gpt-5.3-codex, codex-5.1-mini
/fast mode and reasoning settings do not change the behaviour
Same project and workflow worked normally a couple of days ago
Reinstalling the CLI and VS Code extension did not help