I have been using the /v1/models endpoints for a while and it used to return all available models, including my fine tuned models. But now I don’t get my latest fine tunings (4 models). Is there a max on the returned number of models or… !?
It’s worth to mention that the missing 4 models from the list are still valid to use.
1 Like
Nope. No pagination or even parameters AFAIK.
Maybe give it a couple days and see if it persists. There’s a lot of glitches going on as they move things around and get it setup for the new features.
What was the base model used for the fine-tuning?
1 Like
The base model was gpt-4o-mini-2024-07-18. Same as I have used as base before.
This is from the Fine Tuning Dashboard:
where the topmost is one of the new ones that has a problem in the models endpoint and the one below is behaving as expected.
I just realized, that the /v1/models/{model} returns this for these two models:
The funny thing is, that the “invalid model” is listed and selectable in the Playground!