Allow setting reasoning effort via environment variable

Confirm this is a feature request for the Python library and not the underlying OpenAI API.

• This is a feature request for the Python library

Describe the feature or improvement you're requesting

Problem
If you don’t want the default medium reasoning effort, you currently have to pass reasoning={"effort": ...} on every request or build a wrapper, there’s no simple, per-environment way to set a different default.

Proposal

• Support an env var OPENAI_REASONING_EFFORT with values: minimal, low, medium, high.
• Precedence: per-request param > env var > SDK default.

Example

export OPENAI_REASONING_EFFORT=low

from openai import OpenAI
client = OpenAI()
resp = client.responses.create(model="gpt-5", input=[{"role":"user","content":"Hi"}])
# uses "low" unless explicitly overridden

Notes

• Invalid values: warn once and ignore.
• No breaking changes; opt-in only.
• Helps reduce boilerplate and enables ops-friendly config.

Additional context

No response

Start at the OpenAI client and the responses.create entry point, then trace how reasoning options are resolved before requests are sent. Check existing environment-variable handling and related tests; done means per-request values override OPENAI_REASONING_EFFORT, valid environment values override the SDK default, and invalid values are warned about once and ignored.