`summary="auto"` produces no reasoning summary events on `gpt-5.3-codex` (Responses API, streaming)

[NivAm12]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

SDK: openai-python (latest)
Endpoint: client.responses.create(..., stream=True)
Model: gpt-5.3-codex (also reproduced on gpt-5.1-codex, gpt-5.1-codex-max)

When calling the Responses API with:

reasoning={"effort": "high", "summary": "auto"}
stream=True
the codex variants never emit response.reasoning_summary_part.added / response.reasoning_summary_text.delta / response.reasoning_summary_part.done events during code-generation
prompts. The stream goes straight from request to response.output_text.delta with no summary in between.

The exact same calling code on gpt-5-mini emits reasoning summary events reliably.

### To Reproduce

reasoning={"effort": "high", "summary": "auto"}

### Code snippets

```Python

OS

windows

Python version

Python 3.14.2

Library version

2.21.0