Setting reasoning_effort with the new gpt-5-search-api ChatCompletions model?

[codeofdusk]

Hello,

When I set the reasoning_effort parameter in requests to gpt-5-search-api via the ChatCompletions endpoint:

oa = openai.OpenAI()
oa.chat.completions.create(
    model="gpt-5-search-api",
    reasoning_effort="high",
    messages=[
        {"role": "user", "content": "Hi!"}
    ]
)

I get:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Users\codeofdusk\AppData\Local\Programs\Python\Python311\Lib\site-packages\openai\_utils\_utils.py", line 286, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\codeofdusk\AppData\Local\Programs\Python\Python311\Lib\site-packages\openai\resources\chat\completions\completions.py", line 1156, in create
    return self._post(
           ^^^^^^^^^^^
  File "C:\Users\codeofdusk\AppData\Local\Programs\Python\Python311\Lib\site-packages\openai\_base_client.py", line 1259, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\codeofdusk\AppData\Local\Programs\Python\Python311\Lib\site-packages\openai\_base_client.py", line 1047, in request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': 'Unrecognized request argument supplied: reasoning_effort', 'type': 'invalid_request_error', 'param': None, 'code': None}}

Which seems like a bug (this works fine in responses, using reasoning={"effort": "high"}). Is there an alternative way to set this for the search-capable model?

Thanks in advance!

[_j]

gpt-5-search-api-2025-10-14 will refuse a verbosity parameter and a reasoning_effort parameter, along with any sampling parameters, even when calling directly.

Link to Playground, but you must have set the mode to ‘Chat Completions’ first, as it seems the endpoint setting, only in a menu, is a local remembered setting, not driven by a URL query or any useful pattern: https://platform.openai.com/chat/edit?models=gpt-5-search-api-2025-10-14

This model seems an update of the gpt-4o-search-preview model .. simply providing an instruction-ignoring search engine product with snippets and links. It is a substitute for Chat Completions not having internal tools.

You not being able to send ANY parameters to it is not all that unexpected, because it is not a model that produces anything but Google-like links with referrer utm tracking links for you to click on with a snippet of information.

Fee: $0.01 + search content tokens billed at model rates.

Actual use at defaults:
-$0.019 - 16000 input tokens
+$0.01 - fee
+$0.006 - 565 output tokens

---

Example you pay for:

Example of the competition: