Organization rate limit bug: completion internal model names with "not available" showing

Platform site location for an org that has a fault with model limits:

https://platform.openai.com/settings/organization/limits

There are two categories, the first is named “models in use” - which might be based on recent calling history.

Issue

The listing has non-callable, seemingly internal model names for completions AI models:

babbage-002 also appear later as "babbage:2023-07-21-v2"

More importantly:

  • the correct model ID is not shown anywhere;
  • despite picking my own org tier and not another, “Not available for this tier” is shown.

The models and completions are in service (not shutoff for a few more months):

An organization must have the truth of their organization there, not a generalized listing. There have been odd limits on some models different from the “all other models” catch-all, unique to an organization (unless this has been completely addressed in conjunction with the replacement UI).

Switching to another tier to view its limits does not fix the model ID fault, nor does switching back fix mine.

The only workaround to find the limit is API call x- headers from usage.

Thank you for the detailed report, @_j. I was able to reproduce the issue during testing and have shared your findings with the team for further investigation.