A streaming response is delivered in small pieces while an AI is still producing it. Instead of waiting for the whole answer, a person may see the first words appear and begin reading sooner.
It is like watching a message being typed rather than receiving the complete message all at once. This can make an app feel more responsive, especially when an answer takes a little time.
The first visible words may not be the complete answer. A stream can still be interrupted or end with an error, so an app should make it clear when the response is finished.