Abrupt enforcement of "allowed models"?

I’ve been using these APIs for many many months, and I’ve never had any “allowed models” checked in the interface. (Honestly, I didn’t even know that was a configuration option. But presumably no models checked under either “blocked” or “allowed” would suggested that all are allowed?) Anyway, suddenly tonight I began receiving this error:

Error code: 404 - {'error': {'message': 'The model gpt-4o-2024-11-20 does not exist or you do not have access to it.', 'type': 'invalid_request_error', 'param': None, 'code': 'model_not_found'}}

Nothing in my code had been changed recently. Nothing in my OpenAI account/project configuration had changed. None of my API keys changed.

Did this “allowed model” enforcement logic suddenly change?

And for those experiencing the same thing, I enabled this model and it began working again.

1 Like

That’s indeed odd.

The only particular thing that I found is that if you go into a project, go to project limits, and pick “select models”, you are automatically defaulted to an “allow” view, but with none of the models selected as “allow”.

A “save” with no other action could shut everything off. It should start with “block” and nothing blocked.

I’ve not been hit, employing that model with a project key.