I think we can take the harsh truth 
1 Like
Does anyone have this working in the actual API rather than the playground? There have been mismatches in capability between playground and the actual API in the past.
Here is OpenAI support confirming that response_format is not yet supported for gpt-4.1.
2 Likes
Wait Jeferson and I had planned a vacation together 
Alright, got it. Thanks. Last time I rely on OpenAI Support.
2 Likes
Well, actually you are also the only real human in this forum. 
2 Likes
I thought I had decided not to tell him that
2 Likes
Yes you are right I did that. I am sorry.
2 Likes
Thanks for this list, truly hope OpenAI at some point gives more metadata in the model listing (or provides an API that does) at some point.
1 Like
I got {enabled:true}, after seeing this appear in “security” tab for an organization.
It is next to domain allowlist for chatkit.
Take a look:
https://platform.openai.com/settings/organization/security/ip-allowlist
Now, as there is no announcement about broader availability, the danger is that if it is an API “oops”, and the platform site value changes back to “false” and platform web code depends on it, then you’d have a whitelist, probably still in operation, that can no longer be managed.
Since it is per organization and not per-project, there is no application granularity where you can try it on your API app key, and then “oh well, new project” if IP allowlist access gets screwed up.
Then ponder the offering behind “onboard_apps_dashboard_enabled”