Add support for gpt-4o, gpt-4o-mini, text-embedding-3-small and text-embedding-3-large

[mbarretol]

This pull request adds support for the new gpt-4o-mini model and closes the issue #111, #114 and #115.

[Pederzh]

Thanks for adding the new gpt-4o-mini model, since for each model is also handled the "versioned" name, I've added it.

[danrasmuson]

Excited for this to merged!

[theoremonemehdi]

What is the ETA for this PR to be merged?
Thanks!

[Arit143]

+1 for merging this soon

[Arit143]

hey @dqbd , can this be reviewed, merged and released please? We are quite eager to try out the new gpt4o mini mostly because of the cost efficiency and we are dependant on this library to safely calculate the tokens utilized in case of streams for chat completions.

[rynti]

FYI for anyone eager to just try it out ASAP - as you can see in the implementation of this PR, the underlying token counter seems to be the same as with gpt-4o. Therefore you can probably just pass gpt-4o as the model name instead of gpt-4o-mini.

[theoremonemehdi]

@rynti that's exactly why merging this PR is no brainer. I think everybody is already doing something like model.startsWith('gpt-4o') ? 'gpt-4o' : model

[alsotang]

@rynti that's exactly why merging this PR is no brainer. I think everybody is already doing something like model.startsWith('gpt-4o') ? 'gpt-4o' : model

This method may work currently, but it is not guaranteed. Just like the notorious if condition for determining Windows 9x: if(version.StartsWith("Windows 9"))

[huytool157]

@dqbd could you please take a look and cut a new release if the PR is good?

[breesetposh]

bump, can we get this in here? we're going to have to start forking to move this along, but would love to stay on the main release branch

[huytool157]

Since package author has been unresponsive, I think we should include gpt-4o-2024-08-06 as well so that we can just release all at once

[dqbd]

Hello! Sorry about the long delay! Will take a look as soon as possible!