Is there a way to access reasoning_content when calling Runner.run?

[101scholar]

In most of the cases, it takes time to orchestrate agents. Users are impatient. Is there a way to access reasoning_content so that it can be printed or logged?

An ideal way is to add hooks. However, there seems to be no such hook at the moment. Only final_output is provided when agent ends.

[pakrym-oai]

Reasoning events are available as raw events in the event stream:

  agent = Agent(
        name="Joker",
        model="o3",
        model_settings=ModelSettings(
            reasoning=Reasoning(
                effort="medium",
                summary="auto",
            ),
        ),
        instructions="First call the `how_many_jokes` tool then take a square root of the result and multiply it by 3, then tell that many jokes.",
        tools=[how_many_jokes],
    )

    result = Runner.run_streamed(
        agent,
        input="Hello",
    )
    async for event in result.stream_events():
        # We'll ignore the raw responses event deltas
        if event.type == "raw_response_event":
            if event.data.type == "response.reasoning_summary_text.delta":
                print(event.data.delta, end="", flush=True)
            continue

[101scholar]

I see. I’ll try using run_streamed as a temporary workaround for now. I’d really appreciate it if run could natively support this feature in the future.

[101scholar]

BTW, this only supports reasoning_summary of OpenAI's o-series model. I mean run_stream does not capture reasoning_content from litellm model.

[github-actions]

This issue is stale because it has been open for 7 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 3 days since being marked as stale.