Hello Paul, by any chance do you have an update on this? Lot’s of stuff just hung up in our process due to unavailability of these fine tuned models.
1 Like
Just confirming that I am seeing the same issue.
I have fine-tuned models from two weeks ago that are still available and working fine in the Assistants API.
But two new models I fine-tuned today are not working in the Assistants API. When I try to set the fine-tuned model to my Assistant via the API (https://platform.openai.com/docs/api-reference/assistants/modifyAssistant) , I get a 400 error.
These same models are also not showing up in the dropdown selector in the Assistants web dashboard (https://platform.openai.com/assistants). However, my old fine-tuned models are still showing up there just fine.
Edit - I can also confirm that the models that work have the message (amongst others) “Evaluating model against our usage policies before enabling”, while the newer jobs DO NOT. This definitely seems like the issue.
6 Likes
same problem no response from anyone so far. At first I was surprised to find this. So many people are facing this issue it looks like, need some message from the team to resolve this.
Same problem. No response from any of my reports. Total silence.
3 Likes
This thread has been in-line with my general experience with large system-breaking bugs in the OpenAI API. This forum is mostly just for group moral support, and doesn’t actually seem to be monitored by OpenAI themselves. I do hope that whenever someone notices this is inevitably fixed that they will update the thread here.
3 Likes
@raf @logankilpatrick @atty-openai @sps @PaulBellow hey guys sorry for the huge ping, by any chance do you have an update on this?
1 Like
Hopefully should have an official update soon. It’s on their radar.
Thanks for being patient.
6 Likes
Thanks for flagging everyone! We’re looking into this issue, I’ll update here once there’s more info!
8 Likes
Thank you so much for the update @PaulBellow @gokulraya really appreciate it.
3 Likes
I can confirm that is still happening. This is an ongoing issue and I´ve been trying to reach OpenAI support about this and still no answer. Hope it gets fixed soon.
1 Like
Yeah many of us are facing this since past 2 weeks. Finally yesterday OpenAI staff commented on it and confirmed they’re working on it. I think it will be resolved soon.
2 Likes
@gokulraya Sorry to be a bother, but this is a really big blocker for my team.
Do we have any updates or ETA’s, even just a general one, for the resolution of this bug?
8 Likes
Hello. As of 10/07/2024, fine-tuned models are still not appearing in the Playground or API.
This is a critical issue for my project.
I hope we get an estimated datetime for the fix.
2 Likes
@gokulraya @PaulBellow Asking again, another week has gone by.
Do we have any updates or ETA’s for the resolution of this bug?
As a fellow engineer, I understand/imagine that this isn’t an easy thing to fix, but the least you could do for us is communicate what’s going on, please.
P.S. We followed the support teams ‘turn it off & on again’ solution (generate a new API key & replace the current one on the fine-tuned model), which (as expected) did not resolve the issue.
1 Like
Throwing my support in here too: this issue is blocking a major release of our product built on top of the Assistants API.
1 Like
good to know its not me. Just spent about 8 hours on debugging. Moving on and will monitor this post
Looks like the issue has been resolved. I just checked the list of available models for my Assistant, and I can now see the latest fine-tuned models that were previously missing. Hopefully, everything remains stable moving forward.
3 Likes
Thanks for sharing, confirming the same on my end (in playground at least)!
Kind of disappointing that OpenAI, being valued at 157 billion, cant figure out how to communicate that effectively.
1 Like
Experiencing the same issue trying to select a previously finetuned model as the base model for a new fine tune job, replying to boost visibility
1 Like