Skip to main content
WSS
wss://api.inworld.ai
/
api
/
v1
/
realtime
/
session
Messages
Authorization
type:httpApiKey

Your authentication credentials. For Basic authentication, please populate Basic $INWORLD_RUNTIME_BASE64_CREDENTIAL

Session Update
type:object

Configure session parameters after connecting.

Input Audio Chunk
type:object

Stream audio data to the server using the configured input format.

Commit Audio Buffer
type:object

Tell the server that audio input is ready to process.

Clear Audio Buffer
type:object

Discard buffered audio without processing.

Create Conversation Item
type:object

Send user text, assistant overrides, or tool outputs as conversation items.

Truncate Conversation Item
type:object

Trim an assistant audio item at a specific timestamp.

Delete Conversation Item
type:object

Remove an item from context to manage memory or privacy.

Retrieve Conversation Item
type:object

Fetch a full conversation item by ID.

Create Response
type:object

Trigger response generation with optional modality overrides.

Cancel Response
type:object

Stop the currently streaming response.

Session Created
type:object

Delivered immediately after connection with the default configuration.

Session Updated
type:object

Acknowledges configuration changes and refreshes expiry info.

Conversation Item Added
type:object

Sent whenever the server appends a conversation item.

Conversation Item Done
type:object

Marks when processing for an item is complete.

Response Created
type:object

Confirms a response is in progress.

Response Output Item Added
type:object

Announces new assistant output items.

Response Output Item Done
type:object

Marks completion of an output item.

Response Content Part Added
type:object

Streaming chunks for multi-part responses.

Response Content Part Done
type:object

Signals completion of a content part.

Response Text Delta
type:object

Streaming text token delta.

Response Text Done
type:object

Final text for a content part.

Response Audio Delta
type:object

Streaming audio chunk using the configured output format.

Response Audio Done
type:object

Indicates an audio part finished streaming.

Audio Transcript Delta
type:object

Streaming transcript for generated audio.

Audio Transcript Done
type:object

Final transcript for generated audio.

Function Call Arguments Delta
type:object

Streaming tool-call arguments for function calls.

Function Call Arguments Done
type:object

Complete tool-call arguments payload.

Response Done
type:object

Sent whenever a response finishes streaming.

Speech Started
type:object

Server-side VAD detected speech start.

Speech Stopped
type:object

Server-side VAD detected speech end.

Audio Buffer Committed
type:object

Server committed buffered audio into a conversation item.

Audio Buffer Cleared
type:object

Confirmation that the audio buffer was cleared.

Input Audio Transcription Delta
type:object

Streaming transcription for user audio.

Input Audio Transcription Completed
type:object

Full transcription for a user audio item.

Conversation Item Retrieved
type:object

Response to conversation.item.retrieve.

Conversation Item Truncated
type:object

Indicates the server truncated an assistant audio item.

Conversation Item Deleted
type:object

Confirms deletion of a conversation item.

Error
type:object

Fatal or recoverable errors surfaced on the stream.

Rate Limits Updated
type:object

Coming soon. This event is defined for forward compatibility and is not currently emitted by the Realtime WebSocket service.