Run Status Stucks in Assistant API

Hello everyone. I am facing with a issue about run status and I saw that couple of other people faced with this problem but I couldn’t find any solution. As title says, when I interact with the api it stucks in run status for a while and then it just stays like frozen. This is not happening only with api but also when I test it in the playground. I uploaded screenshot of how it stucks in the playground.

Do you know any solution abot this? I wonder if I am doing something wrong or is it a just bug with assistant api. As I can see, I am not doing anything wrong because just 6 days ago code was working and now it doesnt. It is the same code and also it stucks on playgorund too.

1 Like

I haven’t tested it yet, I’m waiting for an opportunity to test it.

You example screen shot has the" run completed." So its not stuck?

That’s another problem. It shouldn’t be stucked but as you can see when I try to write a new conversation, “Run” button is grey, that means I cant click on “Run” and send the new conversation because actually run is not completed.

Also it burns too much tokens for answers and I dont know why. Just 6 days before today, it was burning much less than current ratio.

I’m running into the same problem as you. In the playground and while going through my python application where I’m stuck in client.beta.threads.runs.retrieve which never returns.
Another day of high unreliability. I’m happy I’m not trying to run this with actual customers on a commercial production system.

I was going to run assistant api on my website commercially but there are a lot of bugs that as developers we cant intervene. I am trying to write a logic about how can I find a way to reset the run because I just find out that when I write another message to assistant like, “continue” it gives me the answer. But What you said is true. It is not reliable if we aim to use the assistant commercially

Guys I think they fixed the bug. I dont know for sure but I am not having the same problem for 5 minutes.

1 Like

For me everything is normal with no problems today

I had this problem all the day until 3 hours ago. After some time problem just disappeared. I still added a logic to create a new run when it stucks though just in case.

1 Like

I’m facing the assistants api run states issue with gpt-4-1106-preview model. The run locks in “queued” forever and the same happens with every other thread I create. Same when I try to cancel some run, it locks in “canceling” forever. I even tried to create another assistant with the 3.5 model but doesn’t seem to have any effect. Anyone have any clue whats going on?

1 Like

This is a bug apperantly and not related to us. Until openai dev team finds a way to fix it we have noting to do to be honest. If creating another run is not working for you(it sometimes fixes the problem for me) you can try to implement another ai api until the problem is fixed. Like I said this is not a cointnous problem, after couple of hours problem goes away.


Did you try switching to the 0125 model (which is the newer version?)