How to integrate the new GPT-5s reasoning_effort parameter in OpenAIChatCompletionClient?

[sourabhdattawad]

I am wondering how to integrate the reasoning_effort parameter into the existing OpenAIChatCompletionClient. Should I wait for a new release, or is it possible with the current implementation? I checked the documentation, but couldn't find any help. Appreciate any leads on this topic. Thanks!

https://openai.com/index/introducing-gpt-5-for-developers/

[ravneetgrewal]

I am waiting for this too - GPT-5 thinking seems to be taking too much time for first token, and passing a

"reasoning":{"effort":"minimal"},

in create method throws a error:

AsyncCompletions.create() got an unexpected keyword argument 'reasoning'

[sourabhdattawad]

@ravneetgrewal or anyone else curious (since this post got a few upvotes):

After looking into the code, I found that you can pass reasoning_effort directly as a parameter to the OpenAIChatCompletionClient.

So instead of doing this:

"reasoning": {"effort": "minimal"}

you should pass it like this:

reasoning_effort = "minimal"

[EShener]

这是来自QQ邮箱的假期自动回复邮件。   您好，我最近正在休假中，无法亲自回复您的邮件。我将在假期结束后，尽快给您回复。