Figure 19.4: Streaming changes the wait you feel, not the wait itself
Figure 19.4 Streaming changes the wait you feel, not the wait itself. Both runs take the
same total time, but the buffered run is a long silent wait ended by the whole answer
at once, while the streamed run puts the first token (in accent) on the screen almost
immediately and then fills in steadily. It front-loads feedback without making the
model any faster.
