Every time I try to use GPT-5 Pro it hangs on the 3 little dots, the model never actually starts thinking nor do anything. Other models, including other GPT variants such as GPT-5 and GPT-5-Codex work fine.
Steps to Reproduce
Try to use GPT-5 Pro model in the lastest cursor IDE
Expected Behavior
The model should respond or at least start doing something.
Operating System
MacOS
Current Cursor Version (Menu → About Cursor → Copy)
Oh so after I tried so send it “Hello” again it eventually responded after a long waiting time. I wonder if it’s just a queuing issue because there is too much demand for this model ?
I tried again with a more complex prompt in Plan mode, it’s been pending for about a minute now : db51b411-9f38-48d4-a5e3-814fe54182dd
Okay, it provided an answer after about 2 minutes of waiting (not thinking). I am sorry for the false positive, it is likely just a load issue. However I wonder, is it just me or do other people experience this too ?
It would be nice though to get this specific model wait time more responsive. For now it seems stuck and if not for this thread I would just assume it’s broken. Also it doesn’t seem to output any thinking traces
@Bitdrifter , could you, maybe reopen it as a bug, because it’s definitely a bug that it doesn’t show neither actual status (e.g. thinking), nor thinking trace, nothing.
Hey, thanks for the report. It looks like GPT-5 Pro is working, but queue times are much longer (1–2+ minutes) before it responds. Since requests do complete, this is likely due to high demand for the model.
You’ve identified an important UX issue: GPT-5 Pro doesn’t show any status indicator or thinking trace while waiting, which makes it look stuck.
Note: some models intentionally don’t show thinking traces, but the absence of any status (for example, “Planning next moves” or “Read” or etc) is definitely a problem.
A couple of questions:
Are you still consistently seeing these long waits?
Do other models (GPT-5, GPT-5-Codex) respond noticeably faster for you now?
I’ll flag this with the team to improve status feedback for GPT-5 Pro so you can see when it’s queued vs. processing.
GPT 5 Pro MAX is impossible to use on Cursor. Broken or NERFED, something is horribly wrong right now.
Started a fresh session and typed: “Hi GPT are you there?”. It took 10 min to get back with “Yes, I am here. How can I help?” and consumed 8.7% of the 272K context. Insane! What on earth is going on?
_____________________________
Steps to Reproduce
Use
Expected Behavior
Work
Operating System
MacOS
Current Cursor Version (Menu → About Cursor → Copy)
Hey, thanks for the message. It looks like GPT-5 Pro is working, but the queue time before a response has gotten much longer (~10 minutes). Since requests do complete, the likely cause is high demand for the model.
FYI: GPT-5 Pro doesn’t show a status indicator or a thinking trace while waiting, which makes it look stuck.
Note: some models intentionally don’t show thinking traces, but the absence of any status (for example, “Planning next moves” or “Read”) is definitely a problem.
I had no output after a bit over 1h of waiting, but it showed that all convo tokens were consumed. I then restarted as nothing was happening -all tokens were consumed with no output and after the restart, the only way to get it working was to change the model. Once changed, it worked as it should and coded a bit in the evening. Later on, I had a message that my credits will expire in 28 days (with the current consumption). I tried again GPT 5 Pro Max before the bed time and the story with consuming tokens while there was no output repeated. I decided to leave it overnight to “think” and hopefully have an output but in the morning there was still no output while it used all tokens. So, I changed the model to Grok and a few lines of code I got a new message that it reached the limit of this month! Reached the limit in just 1 day without coding or benefiting from my subscription payment, this is not ok!!! Something is definitely wrong here!!! Please advise.
