Bad Request

Resolved 💬 7 comments Opened Feb 21, 2026 by ryanc-cf Closed Feb 23, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

This is becoming a serious problem. Codex is nearly unusable.

It's on every request/reply, literally :/

<img width="3796" height="1642" alt="Image" src="https://github.com/user-attachments/assets/45e09641-2fd0-4752-9c41-2f90065932f2" />

View original on GitHub ↗

7 Comments

github-actions[bot] contributor · 4 months ago

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

  • #12119
  • #12055
  • #10571
  • #10780

Powered by Codex Action

etraut-openai contributor · 4 months ago

@ryanc-cf, I won't be able to help diagnose why you're seeing this issue from screenshots. Please use the /feedback slash command to upload your logs and session details, then post your thread ID here.

ryanc-cf · 4 months ago

@etraut-openai I did that, I used /feedback, and went to the link which was straight to this thread. Heck you can even see me using /feedback in the screenshot lol. Waaaaaaaaaaait a min. Am I talking to a robot? :D HEre is the ID: 019c818a-1a3a-71d0-b09f-db01243c933c

etraut-openai contributor · 4 months ago

The thread ID is the part I needed. Unfortunately, I don't see that thread ID in our db, which means that your feedback logs didn't upload successfully. Are you experiencing general network connectivity problems?

We have received some reports of this "Bad request" error recently, but only by a few users — not enough for us to really get a good handle on what causes it. My working theory is that it's correlated to very large tool call outputs. Are you seeing this with only one session, or does it happen for all of your sessions?

ryanc-cf · 4 months ago

It's interesting that you bring up the network issue part because I actually had the SAME thought. So while these 3 requests failed, guess what I had running in the background...... I had a continuous ping running and I got no dropped packets :( So it isn't my connection but that isn't to say it still couldnt be a network hop issue or some other type of network related thing. It's happening with all sessions, I tried the whole make a new session thing.

I also asked the robot itself what it thought the cause was and it seemed to think it was to large of a call to a tool and it said it would break them into smaller pieces, but even that didnt help. The very next line would usually be the bad request error. So it thought the same thing but it was unable to resolve it on it's own by breaking the job up.

ryanc-cf · 4 months ago

@etraut-openai Here is an example of when it suggested the same thing you did after I asked it what it thought the problem was (before I made this ticket).

<img width="3840" height="692" alt="Image" src="https://github.com/user-attachments/assets/da049419-9364-423b-ae24-e1f2c380f544" />

The task I have it working on is simple SASS/CSS and comparing it to a known trusted source with playwright via screenshots. So whenever it makes a change, it checks the source of truth to see if it looks right vs what the site actually shows. It's nothing to crazy where it would be dealing with a lot of data unless a SS is considered such.

etraut-openai contributor · 4 months ago

This appears to be correlated to unusually long sessions and/or user prompts. We're continuing to investigate.

Closing as duplicate of https://github.com/openai/codex/issues/10571.