API for additional model details?

[Draque]

I’ve got a question about pulling info on the various models that you can access via API.

In my application, I’ve got a dropdown that allows users to select from GPT models that are available to them, given their access token and organization. I am able to pull all of these via the endpoint GET api.openai .com/v1/models

The details provided in the JSON that come back are pretty sparse, though. Not all of the models can be accessed via the same endpoint, it doesn’t specify the maximum token count for the models, etc. Is there any place that I can programmatically query that will give me these details? Any assistance here appreciated.

(NOTE: I am aware of GET get api.openai .com/v1/models/{model} as an endpoint, but that gives identical information to the base call, just filtered to a single model.)

[_j]

Here’s the current state of such, which I replied four days ago:

Add a token limit attribute on api.openai.com/v1/models

This is something that is also an unanswered request for features in tiktoken, since it seems openAI doesn’t yet see the value of returning context length per model. Best one can do is dump all the models out and after research, create your own dictionary, also noting the endpoints and formatting methods supported if you want a “pick any model” type application. The best one might do is add some heuristics to figure out what “gpt-3.6-boost-24k” in the model list would mean in the future. Anothe…

TL;DR: you have to write your own, and either probe the models or gather information yourself from documentation.

[PriNova]

As far as I know, it is not possible to programmatically query the token length of a model. I guess you know already the OpenAI model page, where all models are shown with their context length and descriptions?!

[Draque]

I’m aware of that yes. I am looking to futureproof my application though, and avoid making hardcoded assumptions about what models are available and what parameters they work within.