I have a basic understanding of how event streams work. I am currently converting langchain code to directly use OpenAI's API and I have a piece of code I am a bit confused about. def stream_message( messages ): client = OpenAI(api_key=settings.OPENAI_API_KEY) response = client.chat.completions.create(messages=messages, model="gpt-4o", temperature=0.5, stream=True, stop=None) return response response = stream_message(messages,) for sse in response: content = sse_chunk.choices[0].delta.content # some other stuff What I'm confused about is that the python function stream_message is executed once and then is iterated. When I put a breakpoint under the initial API call and inspect it, it gives me the entire response from GPT. Is this really streaming data back? How does this work if I call the function once and I iterate the response value underneath? How is the response getting updated as I iterate the object if it really is streaming a response back? Continue reading...
