A technique where an AI model's response is sent back and displayed incrementally as it's generated, rather than all at once.
Streaming significantly improves perceived responsiveness in chat interfaces, since users see the answer forming immediately instead of waiting for the entire response.
