Streaming

Streaming is a mode where the model sends responses in real-time, token by token, instead of waiting to generate the entire output. This is what makes chats "type" on screen like a human at a keyboard. Beyond the visual effect, streaming has practical value: users start reading immediately, so perceived latency drops, and long responses can be interrupted mid-generation, saving time and costs. In APIs, it's a standard option when building your own applications.