Streaming.
Set stream: true to receive Server-Sent Events in the OpenAI chat.completion.chunk format, ending with data: [DONE].
Live
Streaming is available only when the selected model's catalog details show it is supported and the model is available.
Usage in the stream
Set stream_options.include_usage to true to receive a final chunk with token usage before [DONE].
Disconnects
If you disconnect, the request is cancelled and you are charged only for output already produced, labelled as estimated usage when exact counts are unavailable.