Best Practices for Handling Mid-Stream Errors (Responses API)

API
1

I am using the async client in the following way:

stream = await client.responses.create(
    model="gpt-4.1",
    input=input_messages,
    previous_response_id=previous_response_id,
    stream=True,
)

async for event in stream:  # <-- Error occurs here, mid-stream
    if event.type == 'response.created':
        response_id = event.response.id
    elif event.type == 'response.content_part.added':
        # Push to UI immediately
        send_to_client(event.part.text)
    elif event.type == 'response.output_text.delta':
        send_to_client(event.delta)
    elif event.type == 'response.completed':
        # Success
        pass

Sometimes I get a random error during the for loop. Example of error:
httpx.RemoteProtocolError: peer closed connection without sending complete message body (incomplete chunked read)

My question is what is the best way to handle this?

My stream is already in the middle, I may have already called some tools, sent some tokens to the client, etc.

RemoteProtocolError: peer closed connection without sending complete message body (incomplete chunked read)