Interface RealtimeCapabilities

Whether the model can produce audio output directly.

Whether the model automatically generates a reply after receiving tool results.

Whether function call items already in the chat context can be resumed.

Whether generated assistant messages can be truncated after interruption.

Whether the chat context can be updated mid-session.

Whether the instructions can be updated mid-session.

Whether the tools can be updated mid-session.

Whether the model can synchronize generated transcript timing natively.

Whether the tool and tool choice can be specified per response.

Whether the model emits server-side speech start and stop events for turn taking.

Whether the model emits user audio transcription events.