ted3
November 9, 2023, 1:25am
1
Hey all,
To prevent breakages in generation I’m pegging the specific model used in different features. However, I’d hate to miss a deprecation date as well.
If we could get the deprecation date to be a part of the returned object in the Models endpoint that would be great.
Explore developer resources, tutorials, API docs, and dynamic examples to get the most out of OpenAI's platform.
Basically I’d love to just cache the deprecation dates every once and a while have my system warn if somewhere is using a soon-to-be-deprecated model.
Good idea. Thanks for sharing.
Welcome to the community. I hope you stick around. We’ve got a good place going with a lot of great people.
As an automation tester, I fully agree: deprecation dates should be part of the models endpoint.
_j
March 5, 2026, 9:12am
4
If you want programmatic use, such as displaying a warning in a UI along with selection or use of a particular model, this should keep you going for a while:
Your code can also surface data from this Python dict I made;
some fine-tunings have been allowed past official shutoff,
other deprecation announcements haven’t made any distinction
about their fine-tuning variants.
DEPRECATIONS = {
"gpt-3.5-turbo-instruct"
If you are actively using the models endpoint, unavailable models that are shut off will of course be unavailable.