models

Streaming

A delivery mode where a model sends its output token by token as it generates, rather than waiting until the full response is complete.