Expand description
Direct, typed incremental OpenAI Chat ↔ Responses conversion.
These adapters deliberately keep the source collector and target framing separate. A source event is validated first, then the corresponding target event is emitted; no intermediate JSON or completed source object is made for partial events.
Structs§
- Chat
ToResponses Context - Original Responses request and effective controls retained during request preparation. A stream cannot invent these fields from a Chat chunk.
- Chat
ToResponses Stream - Responses
ToChat Stream - Each push returns only newly emitted target chunks. Source item/part order is retained when concurrent Responses messages feed Chat’s single content field.
- Stream
End - Stream
Limits