Performance
Streaming Tokens
A delivery mode where model output tokens are sent to the client incrementally as they are generated rather than all at once. Streaming makes AI coding tools feel faster and allows users to interrupt unhelpful responses early.