I’m moving a chat feature to the OpenAI Responses API and am unsure how much conversation state I need to manage myself. If I pass a previous_response_id, does the next request have the context it needs from the earlier response, or should my app still resend the full message history?

I’m also trying to understand what happens when a conversation gets long or an earlier response is no longer available. How are other developers handling conversation state in the Responses API without making every request unnecessarily large?