[Feature Request]: Update OpenAI default model to use gpt-4o-mini from gpt-3.5-turbo

[jan-rodriguez]

Feature Description

Update OpenAI default model to use gpt-4o-mini from gpt-3.5-turbo

Reason

No response

Value of Feature

According to OpenAI's documentation they recommend using gpt-4o-mini over gpt-3.5-turbo

as it is cheaper, more capable, multimodal, and just as fast

[logan-markewich]

gpt-4o-mini is also way worse at function calling. Probably changing this default will cause a lot of issues 😓 But will need to happen at some point

[jan-rodriguez]

Really? From my experience it's better at function calling (for my particular case). It's actually the reason I was making this change. I was getting unexpected responses from gpt-3.5-turbo where it would return multiple function calls for a simple task when there were multiple tools specified, but using gpt-4o-mini was returning the correct number of function calls

[jan-rodriguez]

We can also make it gpt-4o instead of gpt-4o-mini as it would do better in function calling, but would be significantly more expensive

[logan-markewich]

I think 4o-mini is the right call, but it may take some significant validation (mostly making sure a lot of our agentic openai notebooks still work as expected)