Response-Items: die Bausteine

Arbeiten mit der OpenAI Responses API

James Chapman

AI Curriculum Manager, DataCamp

Die Reise

Dein Item-Werkzeugkasten

for item in response.output:
    print(item)

ResponseReasoningItem(id='rs_0471305df20125be00692440df2ed88193bd532a15ee6ab7dc',
                      summary=[], type='reasoning', content=None, encrypted_content=None,
                      status=None)

ResponseOutputMessage(id='msg_0471305df20125be00692440df642481938921f6bd2c119688',
                      content=[ResponseOutputText(annotations=[], text='The OpenAI...',
                                                  type='output_text', logprobs=[])],
                      role='assistant', status='incomplete', type='message')

Items als Bausteine

for item in response.output:

    if item.type == "message":

        if item.status == "incomplete":

            print("Message cut-off!")

Message cut-off!

Items als Bausteine

for item in response.outputs:

    if item.type == "reasoning"

        if item.summary:
            print(f"Reasoning Summary: {item.summary[0]}")
        else:
            print("No reasoning summary available.")

    if item.type == "message":
        print(f"Assistant: {item.content[0].output_text}")

No reasoning summary available.

Assistant: The OpenAI...

Als Nächstes ...

Aktuelle Item-Typen

'reasoning'
'message'

for item in response.output:
    if item.type == "message":
        if item.status == "incomplete":
            print("Message cut-off!")

Neue Item-Typen

'web_search_call'
'function_call'
'function_call_output'

Lass uns üben!

Arbeiten mit der OpenAI Responses API