responses.create() hangs indefinitely with GPT-5 models while direct REST API call completes successfully

[pietroperona]

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

When using the Responses API with GPT-5 models (specifically gpt-5-nano-2025-08-07) via the Python SDK, the call client.responses.create() hangs indefinitely and eventually times out after ~30 seconds.
The same request, when made directly via the REST API (curl), completes successfully in about 12 seconds with "status": "completed".

This indicates the issue is likely within the Python SDK implementation, not the OpenAI backend.

To Reproduce

Working REST API call (completes in ~12 seconds)

curl https://api.openai.com/v1/responses \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "model": "gpt-5-nano-2025-08-07",
    "instructions": "Analyze the following newsletters and produce a 400-word abstract with 5 key bullet points.",
    "input": "Sample newsletter content here...",
    "store": true
  }'

Result: Completes in ~12 seconds with "status": "completed" and full output available.

Failing Python SDK call (times out after 30+ seconds)

import asyncio
from openai import AsyncOpenAI

async def test():
    client = AsyncOpenAI(api_key="sk-...")

    response = await client.responses.create(
        model="gpt-5-nano-2025-08-07",
        instructions="Analyze the provided content and summarize it.",
        input="Sample newsletter content here...",
        max_output_tokens=600,
        store=True,
    )

    print(f"Initial response: id={response.id}, status={response.status}")

    for _ in range(30):
        await asyncio.sleep(1)
        response = await client.responses.retrieve(response.id)
        print(f"Status: {response.status}")
        if response.status == "completed":
            print("Completed successfully")
            return

    raise TimeoutError("Timed out after 30 seconds")

asyncio.run(test())

Result: The call times out after 30+ seconds; status never becomes "completed".

Code snippets

OS

14.6

Python version

3.12

Library version

2.6.1