Adjust code to changes announced today

[Niek]

See: https://openai.com/blog/new-models-and-developer-products-announced-at-devday

In particular, new pricing and models.

[MitchBoss]

I gave some of this a crack for you: #341 👍

[TeunvdWeij]

So will we be able to use gpt4-turbo and use images? That would be awesome!

And are plug-ins feasible?

[Niek]

I have updated the prices now to match https://openai.com/pricing#language-models

There is supposed to be a gpt-4-1106-vision-preview model, but I have not seen it yet.

[lazaroagomez]

What about the 128K context? 😶 It's only letting me use 4096 Tokens

[dootss]

What about the 128K context? 😶 It's only letting me use 4096 Tokens

1106 has 128k context token but has a maximum reply output of 4096 as far as i know

[MitchBoss]

What about the 128K context? 😶 It's only letting me use 4096 Tokens

That's just max output - you just need to adjust the token threshold. :)

You are more likely to hit the api limits before you get close to 128k though: Error: 429 - Rate limit reached for gpt-4-1106-preview in organization org-xxx on tokens per min. Limit: 20000 / min. Please try again in 3ms. Visit https://platform.openai.com/account/rate-limits to learn more.

[SetoKaiba]

What about add gpt-4-1106-vision-preview as well?

[Niek]

I don't see gpt-4-1106-vision-preview in the API yet. Not sure if that's because my account has no access to it yet, or if it simply isn't released yet?

[timothelaborie]

I don't see gpt-4-1106-vision-preview in the API yet. Not sure if that's because my account has no access to it yet, or if it simply isn't released yet?

The model is called gpt-4-vision-preview

[davkap92]

Awesome project thanks, yeah the ability to drag an image into the chat and use the vision model would be awesome. https://platform.openai.com/docs/guides/vision?lang=python
I don't know if one can then continue chatting with vision model with context

[Niek]

Ah, so gpt-4-vision-preview can be used to read images, and dall-e-3-* can be used to generate new ones.

This would definitely require some changes, as we need to support upload functionality.

[Ryan526]

Ah, so gpt-4-vision-preview can be used to read images, and dall-e-3-* can be used to generate new ones.

This would definitely require some changes, as we need to support upload functionality.

Is the ability to upload images still something you might plan on adding to this?