Codex insanely slow at 3 chats + sub-agents

What version of the Codex App are you using (From “About Codex” dialog)?

Version 26.623.101652

What subscription do you have?

Codex Pro

What platform is your computer?

Microsoft Windows NT 10.0.26200.0 x64

What issue are you seeing?

Insanely slow and stalling today. I've noticed this usually occurs when I have 3 different threads open. Currently running 2 chats in a main and branch of the same project. The 3rd chat is is on another main branch, separate project. GPT 5.5 Extra High for all three chats.

I've noticed that two chats seems okay, still somewhat slow. But it's common as soon as I'm in three, it's like one of them gets dreadfully slow, locks up and says thinking for 10 minutes and finally responds back on a 20k token run that shouldn't have taken very long.

I noticed that this slow and freezing behavior tends to be owned by a single chat of the three for the majority of the time. I'll even start a new chat but it's almost like that window is getting locked up and it's not related to the specific chat itself.

The sub-agents have also been incredibly slow today too and some never even completed a simple run after 20-30 minutes on a task that should have taken around 8 minutes tops at less than 60k tokens. Normally the sub-agents works great, but today, this has been freezing, the control agent kept telling me it was waiting on the receipt from the sub-agent but then would finally verify after 20 minutes of waiting and me pinging it, then it would investigate. Often it would find the work was completed but not reported back. So i'm not sure what's happening today.

What steps can reproduce the bug?

Open 3 different Codex App Windows in their own window. Maybe try two in the same project with one on main and one on a branch, and a 3rd chat from a different project and set all of them to GPT 5.5 Extra High Standard speed.

Have one of them run the sub-agent and give it a task on a Git Repo Branch and see if it runs properly or gets hung up. I haven't even used up a 5 hour block today and I'm used to doing that between 3 active chats. I haven't even come but halfway today on any of them. I can't, it's too slow to let me. 20-30 min wait times to have to redo a run is bad and waste tokens and my time horribly.

What is the expected behavior?

Normally, they move relatively fast and don't get hung up. At 3+ chats they can get slower, but not like this.

sub-agent definitely should not be getting stuck and not reporting receipts or back to the controller agent. And it definitely should be finishing within 5 minutes on the runs i'm doing but it's around 20-30 minutes at times.

Additional information

Status on OpenAI seemed fine from the website status page. I was getting all green from what I saw. Internet speeds solid and holding. I've seen it tell me it needed to reconnect attempts and retries multiple times even though nothing else I have is acting up.