feature: support openai's reasoning_effort parameter

[Madoshakalaka]

Feature request

Please support the reasoning_effort parameter for openai reasoning models like o3-mini.

as discussed here: https://community.openai.com

by default the reasoning effort is medium

Motivation

reasoning_effort high really seems to improve the coding ability as shown in benchmarks

Other

I want to contribute but I'm not sure how to design the user interface. I don't think we have a UI for model parameters?

[esezen]

@yetone to confirm but my understanding is this is already supported. I tried it locally and logged the parameters and it appears like you can set it at the provider level. I opened this PR to document it. Let me know if this works

[Madoshakalaka]

thanks!

More context on the use case: I toggle between reasoning providers (o3-mini) and non-reasoning ones (claude) depending on the task. Passing reasoning_effort high makes a lot of sense!