Live voice agent event models

These models are referenced by Live client events and server events.

Models

OpenAI.AnswerResource

A completed question always includes its name, including null when unnamed.

Variants by type: "predicate": OpenAI.AnswerResourcePredicate; "choice": OpenAI.AnswerResourceChoice; "score": OpenAI.AnswerResourceScore; "refusal": OpenAI.AnswerResourceRefusal.

Property Type Required Description
type OpenAI.AnswerResourceType Yes

OpenAI.AnswerResourceChoice

All of: OpenAI.AnswerResource.

Property Type Required Description
type "choice" Yes The type of the object. Always choice.
name Any of: string; null Yes minLength: 0.
choice OpenAI.ChoiceValueResource Yes
probabilities Array of OpenAI.ChoiceProbabilityResource Yes minItems: 0. maxItems: 2000.
confidence OpenAI.numeric Yes

OpenAI.AnswerResourcePredicate

All of: OpenAI.AnswerResource.

Property Type Required Description
type "predicate" Yes The type of the object. Always predicate.
name Any of: string; null Yes minLength: 0.
probability OpenAI.numeric Yes

OpenAI.AnswerResourceRefusal

The model declined to answer this question. Other questions in the same request can still receive answers.

All of: OpenAI.AnswerResource.

Property Type Required Description
type "refusal" Yes The type of the object. Always refusal.
name Any of: string; null Yes minLength: 0.

OpenAI.AnswerResourceScore

All of: OpenAI.AnswerResource.

Property Type Required Description
type "score" Yes The type of the object. Always score.
name Any of: string; null Yes minLength: 0.
score OpenAI.numeric Yes
probabilities Array of OpenAI.ScoreProbabilityResource Yes minItems: 0. maxItems: 2000.
confidence OpenAI.numeric Yes

OpenAI.AnswerResourceType

Any of: string, "predicate" or "choice" or "score" or "refusal".

OpenAI.ChoiceProbabilityResource

Property Type Required Description
value OpenAI.ChoiceValueResource Yes
probability OpenAI.numeric Yes

OpenAI.ChoiceValueResource

Choice values are typed: a string and a boolean with the same text are distinct.

One of: string, boolean.

OpenAI.DecisionResponse

Property Type Required Description
model string Yes The model used. minLength: 0.
answers Array of OpenAI.AnswerResource Yes Answers in question order. Any question can return a refusal instead of an answer; other questions can still receive answers. A refusal has type: "refusal" and the question's name, or null if it was unnamed. minItems: 0. maxItems: 2000.
usage OpenAI.ResponseUsageResource Yes Token usage for the request.

OpenAI.InputItem

An item representing part of the context for the response to be generated by the model. Can contain text, images, and audio inputs, as well as previous assistant responses and tool call outputs.

For input item variants, see the Responses API reference.

Property Type Required Description
type OpenAI.InputItemType Yes

OpenAI.InputItemType

Any of: string, "message" or "output_message" or "file_search_call" or "computer_call" or "computer_call_output" or "web_search_call" or "function_call" or "function_call_output" or "tool_search_call" or "tool_search_output" or "additional_tools" or "configuration_update" or "reasoning" or "compaction" or "image_generation_call" or "code_interpreter_call" or "local_shell_call" or "local_shell_call_output" or "shell_call" or "shell_call_output" or "apply_patch_call" or "apply_patch_call_output" or "mcp_list_tools" or "mcp_approval_request" or "mcp_approval_response" or "mcp_call" or "custom_tool_call_output" or "custom_tool_call" or "item_reference".

OpenAI.InputTokensDetailsResource

Property Type Required Description
cached_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.
cache_write_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.

OpenAI.LiveAllowedServerEventParam

A Live server event selector for the WebRTC frontend data channel.

Property Type Required Description
type string Yes The outer Live server event type. Use 'response.event' for Responses events. minLength: 1. maxLength: 256.
response_event string No The nested Responses event type. Required when type is 'response.event'; forbidden for other event types. minLength: 1. maxLength: 256.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveAllowedToolsChoiceParam

Property Type Required Description
type "allowed_tools" Yes The tool choice type. Always allowed_tools. default: "allowed_tools".
tools Array of Any of: OpenAI.LiveFunctionToolChoiceParam; OpenAI.LiveMCPToolChoiceParam; OpenAI.LiveSpecificFileSearchParam; OpenAI.LiveSpecificWebSearchParam; OpenAI.LiveSpecificWebSearchPreviewParam; OpenAI.LiveSpecificImageGenParam; OpenAI.LiveSpecificComputerParam; OpenAI.LiveSpecificCodeInterpreterParam; OpenAI.LiveSpecificProgrammaticToolCallingParam; OpenAI.LiveSpecificFunctionShellParam; OpenAI.LiveSpecificCustomToolParam; OpenAI.LiveSpecificApplyPatchParam Yes The tools that the delegated Responses model may call. minItems: 1. maxItems: 128.
mode Any of: OpenAI.LiveToolChoiceValueEnum; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveApplyPatchToolInputParam

Property Type Required Description
type "apply_patch" Yes default: "apply_patch".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveAudioFormat

Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.

Variants by type: "audio/pcm": OpenAI.LiveSessionAudioFormatPCMParam; "audio/pcmu": OpenAI.LiveSessionAudioFormatPCMUParam; "audio/pcma": OpenAI.LiveSessionAudioFormatPCMAParam.

Property Type Required Description
type OpenAI.LiveAudioFormatType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveAudioFormatType

Any of: string, "audio/pcm" or "audio/pcmu" or "audio/pcma".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveClientConfigParam

Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.

Property Type Required Description
data_channel OpenAI.LiveDataChannelConfigParam Yes Client and server event permissions for the WebRTC frontend data channel.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveClientEvent

Live voice-agent client event. Foundry sends the session start event.

Property Type Required Description
type OpenAI.LiveClientEventType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveClientEventType

Any of: string, "session.start" or "session.update" or "session.input_audio.append" or "session.input_audio.mute" or "session.input_audio.unmute" or "session.output_audio.fixture.play" or "session.instructions.append" or "session.thinking.append" or "session.commentary.append" or "decision.create" or "decision.item.create" or "decision.item.delete" or "response.item.create" or "response.create" or "session.close".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveCodeInterpreterToolInputParam

Property Type Required Description
type "code_interpreter" Yes default: "code_interpreter".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveComputerToolInputParam

Property Type Required Description
type "computer" Yes default: "computer".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveContainerMemoryLimit

Type: "1g" or "4g" or "16g" or "64g".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveContainerNetworkPolicyDisabledParam

Property Type Required Description
type "disabled" Yes Disable outbound network access. Always disabled. default: "disabled".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveContainerReferenceParam

Property Type Required Description
type "container_reference" Yes References a container created with the /v1/containers endpoint default: "container_reference".
container_id string Yes The ID of the referenced container.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveContextWindowUsage

The latest measured context-window usage of the Live model. This is separate from billing usage and delegated Responses token usage.

Property Type Required Description
usage_ratio OpenAI.numeric Yes The latest active context token count divided by the Live model context limit. Can decrease after compaction and may lag between measured audio frames.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveCustomToolInputParam

Property Type Required Description
type "custom" Yes default: "custom".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveCustomVoiceParam

Property Type Required Description
id string Yes minLength: 1. maxLength: 128.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDataChannelConfigParam

Control which Live events an untrusted WebRTC frontend can send and receive over its data channel. These restrictions do not apply to trusted sideband connections.

Property Type Required Description
allowed_client_events Exactly one of: "all"; Array of string No Client event types that the frontend data channel may send. Use 'all' to allow every client event; an empty array allows none. Omission preserves the existing allow-all behavior. default: "all".
allowed_server_events Exactly one of: "all"; Array of OpenAI.LiveAllowedServerEventParam No Server events that may be sent to the frontend data channel. Use 'all' to allow every server event; an empty array allows none. Omission preserves the existing allow-all behavior. Responses events use an object with type 'response.event' and a response_event selector. default: "all".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionChoiceParam

Property Type Required Description
value Exactly one of: string; boolean Yes
description string No

OpenAI.LiveDecisionChoiceQuestionParam

All of: OpenAI.LiveDecisionQuestion.

Property Type Required Description
type "choice" Yes
name string No
instructions Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam; OpenAI.LiveDecisionInputJsonParam Yes
choices Array of OpenAI.LiveDecisionChoiceParam Yes minItems: 2. maxItems: 255.

OpenAI.LiveDecisionCompleted

Property Type Required Description
type "decision.completed" Yes default: "decision.completed".
decision OpenAI.DecisionResponse Yes The native Decision API response, including its model, ordered answers, and token usage.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionCustomToolCallOutputParam

All of: OpenAI.LiveDecisionInputItemParam.

Property Type Required Description
id Any of: string; null No
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "custom_tool_call_output" Yes
output Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam Yes

OpenAI.LiveDecisionCustomToolCallParam

All of: OpenAI.LiveDecisionInputItemParam.

Property Type Required Description
id Any of: string; null No
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "custom_tool_call" Yes The item type. Always custom_tool_call.
name string Yes The name of the custom tool to call. minLength: 1. maxLength: 64. pattern: "^[a-zA-Z0-9_-]+$".
namespace Any of: string; null No
input string Yes The input to the custom tool call.

OpenAI.LiveDecisionFunctionCallOutputParam

All of: OpenAI.LiveDecisionInputItemParam.

Property Type Required Description
id Any of: string; null No
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "function_call_output" Yes The type of the function tool call output. Always function_call_output.
output Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam Yes

OpenAI.LiveDecisionFunctionCallParam

All of: OpenAI.LiveDecisionInputItemParam.

Property Type Required Description
id Any of: string; null No
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "function_call" Yes The item type. Always function_call.
name string Yes minLength: 1. maxLength: 128. pattern: "^[a-zA-Z0-9_.-]+$".
namespace Any of: string; null No
arguments string Yes The function arguments as a JSON string.

OpenAI.LiveDecisionInputImageParam

Property Type Required Description
type "input_image" Yes default: "input_image".
image_url string Yes A base64-encoded image data URL or a publicly accessible HTTP(S) image URL. format: "uri". maxLength: 20971520. pattern: "^(data:|https?://)".
detail Any of: OpenAI.LiveDetailEnum; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionInputItemParam

Variants by type: "message": OpenAI.LiveDecisionInputItemParamMessage; "function_call": OpenAI.LiveDecisionFunctionCallParam; "function_call_output": OpenAI.LiveDecisionFunctionCallOutputParam; "custom_tool_call": OpenAI.LiveDecisionCustomToolCallParam; "custom_tool_call_output": OpenAI.LiveDecisionCustomToolCallOutputParam.

Property Type Required Description
type OpenAI.LiveDecisionInputItemParamType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionInputItemParamMessage

All of: OpenAI.LiveDecisionInputItemParam.

Property Type Required Description
type "message" Yes

OpenAI.LiveDecisionInputItemParamType

Any of: string, "message" or "function_call" or "function_call_output" or "custom_tool_call" or "custom_tool_call_output".

OpenAI.LiveDecisionInputJsonParam

Property Type Required Description
type "input_json" Yes default: "input_json".
data Exactly one of: OpenAI.LiveDecisionJsonObject; Array of OpenAI.LiveDecisionJsonObject Yes

OpenAI.LiveDecisionInputTextParam

Property Type Required Description
type "input_text" Yes default: "input_text".
text string Yes The text input to the model. maxLength: 10485760.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionJsonObject

Additional properties: any.

OpenAI.LiveDecisionPredicateQuestionParam

All of: OpenAI.LiveDecisionQuestion.

Property Type Required Description
type "predicate" Yes
name string No
instructions Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam; OpenAI.LiveDecisionInputJsonParam Yes

OpenAI.LiveDecisionQuestion

Variants by type: "choice": OpenAI.LiveDecisionChoiceQuestionParam; "predicate": OpenAI.LiveDecisionPredicateQuestionParam; "score": OpenAI.LiveDecisionScoreQuestionParam.

Property Type Required Description
type OpenAI.LiveDecisionQuestionType Yes

OpenAI.LiveDecisionQuestionType

Any of: string, "choice" or "predicate" or "score".

OpenAI.LiveDecisionQuestionsInputParam

Questions evaluated using the current transcript and the session's shared Decision items.

Property Type Required Description
questions Array of OpenAI.LiveDecisionQuestion Yes Native Decision API questions evaluated together, in order. minItems: 1. maxItems: 64.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionScoreLevelParam

Property Type Required Description
label string Yes
description string No

OpenAI.LiveDecisionScoreQuestionParam

All of: OpenAI.LiveDecisionQuestion.

Property Type Required Description
type "score" Yes
name string No
instructions Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam; OpenAI.LiveDecisionInputJsonParam Yes
levels Array of OpenAI.LiveDecisionScoreLevelParam Yes minItems: 2. maxItems: 10.

OpenAI.LiveDecisionStoredCustomToolCallOutputParam

All of: OpenAI.LiveDecisionStoredInputItemParam.

Property Type Required Description
id string Yes
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "custom_tool_call_output" Yes
output Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam Yes

OpenAI.LiveDecisionStoredCustomToolCallParam

All of: OpenAI.LiveDecisionStoredInputItemParam.

Property Type Required Description
id string Yes
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "custom_tool_call" Yes The item type. Always custom_tool_call.
name string Yes The name of the custom tool to call. minLength: 1. maxLength: 64. pattern: "^[a-zA-Z0-9_-]+$".
namespace Any of: string; null No
input string Yes The input to the custom tool call.

OpenAI.LiveDecisionStoredFunctionCallOutputParam

All of: OpenAI.LiveDecisionStoredInputItemParam.

Property Type Required Description
id string Yes
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "function_call_output" Yes The type of the function tool call output. Always function_call_output.
output Exactly one of: string; Array of Any of: OpenAI.LiveDecisionInputTextParam; OpenAI.LiveDecisionInputImageParam Yes

OpenAI.LiveDecisionStoredFunctionCallParam

All of: OpenAI.LiveDecisionStoredInputItemParam.

Property Type Required Description
id string Yes
call_id string Yes The unique ID of the function tool call generated by the model. minLength: 1. maxLength: 64.
status Any of: OpenAI.LiveFunctionCallItemStatus; null No
type "function_call" Yes The item type. Always function_call.
name string Yes minLength: 1. maxLength: 128. pattern: "^[a-zA-Z0-9_.-]+$".
namespace Any of: string; null No
arguments string Yes The function arguments as a JSON string.

OpenAI.LiveDecisionStoredInputItemParam

Variants by type: "message": OpenAI.LiveDecisionStoredInputItemParamMessage; "function_call": OpenAI.LiveDecisionStoredFunctionCallParam; "function_call_output": OpenAI.LiveDecisionStoredFunctionCallOutputParam; "custom_tool_call": OpenAI.LiveDecisionStoredCustomToolCallParam; "custom_tool_call_output": OpenAI.LiveDecisionStoredCustomToolCallOutputParam.

Property Type Required Description
type OpenAI.LiveDecisionStoredInputItemParamType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionStoredInputItemParamMessage

All of: OpenAI.LiveDecisionStoredInputItemParam.

Property Type Required Description
type "message" Yes

OpenAI.LiveDecisionStoredInputItemParamType

Any of: string, "message" or "function_call" or "function_call_output" or "custom_tool_call" or "custom_tool_call_output".

OpenAI.LiveDecisionTriggerInputParam

Property Type Required Description
type "decisions" Yes default: "decisions".
event_type "session.input_transcript.delta" or "session.output_transcript.delta" Yes
decisions OpenAI.LiveDecisionQuestionsInputParam Yes Questions evaluated using the current transcript and the session's shared Decision items.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionsDelegationUpdateParam

All of: OpenAI.LiveSessionUpdateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "decisions" Yes
decisions OpenAI.LiveDecisionsSettingsUpdateParam No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionsSettingsInputParam

Property Type Required Description
questions Array of OpenAI.LiveDecisionQuestion Yes Native Decision API questions evaluated together, in order. minItems: 1. maxItems: 64.
model string Yes minLength: 1.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDecisionsSettingsUpdateParam

Property Type Required Description
model string No minLength: 1.
questions Array of OpenAI.LiveDecisionQuestion No Native Decision API questions evaluated together, in order. minItems: 1. maxItems: 64.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDelegationReasoningInputParam

Reasoning options for Responses requests made on behalf of the Live session.

Property Type Required Description
effort Any of: OpenAI.LiveReasoningEffort; null No
summary Any of: OpenAI.LiveReasoningSummary; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDelegationTextInputParam

Text generation options for the Live session’s Responses backend.

Property Type Required Description
verbosity Any of: OpenAI.LiveTextVerbosity; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveDetailEnum

Type: "low" or "high" or "auto" or "original".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveFileSearchToolInputParam

Property Type Required Description
type "file_search" Yes default: "file_search".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveFunctionCallItemStatus

Type: "in_progress" or "completed" or "incomplete".

OpenAI.LiveFunctionToolChoiceParam

Property Type Required Description
type "function" Yes The tool to call. Always function. default: "function".
name string Yes The name of the function tool to call.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveFunctionToolInputParam

A function tool available to the Responses backend when the Live model delegates a task.

Property Type Required Description
type "function" Yes The tool type. Always function. default: "function".
name string Yes The name the delegated Responses model uses when calling this function.
description Any of: string; null No
parameters Any of: object with additional values of any; null No
strict Any of: boolean; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveHostedShellContainerAutoParam

Property Type Required Description
type "container_auto" Yes Automatically creates a container for this request default: "container_auto".
file_ids Any of: Array of string; null No
memory_limit Any of: OpenAI.LiveContainerMemoryLimit; null No
network_policy Any of: OpenAI.LiveContainerNetworkPolicyDisabledParam; OpenAI.LiveHostedShellNetworkPolicyAllowlistParam; null No
skills Any of: Array of Any of: OpenAI.LiveSkillReferenceParam; OpenAI.LiveInlineSkillParam; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveHostedShellNetworkPolicyAllowlistParam

Property Type Required Description
type "allowlist" Yes Allow outbound network access only to specified domains. Always allowlist. default: "allowlist".
allowed_domains Array of string Yes A list of allowed domains when type is allowlist. minItems: 1.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveHostedShellToolInputParam

A Responses shell tool. Use a hosted container or return local shell results with response.item.create. Domain secrets are not supported.

Property Type Required Description
type "shell" Yes default: "shell".
environment Any of: OpenAI.LiveHostedShellContainerAutoParam; OpenAI.LiveContainerReferenceParam; OpenAI.LiveLocalEnvironmentParam; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveImageGenerationToolInputParam

Property Type Required Description
type "image_generation" Yes default: "image_generation".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialAssistantMessageItemParam

An assistant message included in the initial text history of a Live session.

All of: OpenAI.LiveInitialItem.

Property Type Required Description
id Any of: string; null No
type "message" No The history item type. Always message. default: "message".
status Any of: OpenAI.LiveInitialMessageStatus; null No
role "assistant" Yes The author of this history message. Always assistant.
content Array of Any of: OpenAI.LiveInitialTextContentPartParam; OpenAI.LiveInitialOutputTextContentPartParam Yes The message content. Supply exactly one text part for the initial Live conversation history. minItems: 1. maxItems: 1.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialDeveloperMessageItemParam

A developer message included in the initial text history of a Live session.

All of: OpenAI.LiveInitialItem.

Property Type Required Description
id Any of: string; null No
type "message" No The history item type. Always message. default: "message".
status Any of: OpenAI.LiveInitialMessageStatus; null No
role "developer" Yes The author of this history message. Always developer.
content Array of OpenAI.LiveInitialInputTextContentPartParam Yes The message content. Supply exactly one text part for the initial Live conversation history. minItems: 1. maxItems: 1.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialInputTextContentPartParam

Text supplied in a developer or user message when starting a Live session.

Property Type Required Description
type "input_text" No The text content type. Always input_text. default: "input_text".
text string Yes The message text to include in the Live session’s initial conversation history.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialItem

A developer, user, or assistant message supplied as text history before the Live session starts.

Variants by role: "developer": OpenAI.LiveInitialDeveloperMessageItemParam; "user": OpenAI.LiveInitialUserMessageItemParam; "assistant": OpenAI.LiveInitialAssistantMessageItemParam.

Property Type Required Description
role OpenAI.LiveInitialItemType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialItemType

Any of: string, "developer" or "user" or "assistant".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialMessageStatus

Type: "incomplete" or "completed".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialOutputTextContentPartParam

Assistant output text supplied as conversation history when starting a Live session.

Property Type Required Description
type "output_text" Yes The text content type. Always output_text. default: "output_text".
text string Yes The message text to include in the Live session’s initial conversation history.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialSessionAudioOutputParam

Settings for speech generated by the Live model. Choose the voice before starting the session.

Property Type Required Description
voice Exactly one of: string; "alloy" or "ash" or "ballad" or "beacon" or "bossa" or "brise" or "cedar" or "cinder" or "coral" or "delta" or "echo" or "flitz" or "gleam" or "harema" or "juni" or "marin" or "meridian" or "nira" or "noeul" or "nuri" or "quartz" or "ripple" or "sage" or "shida" or "shimmer" or "sillage" or "stone" or "tempo" or "verse" or "vesper" or "willow"; OpenAI.LiveCustomVoiceParam No The voice used for Live speech, as a built-in voice name or a custom voice object containing its ID. Defaults to marin and cannot change after startup.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialSessionAudioParam

Startup audio configuration. Only primary WebSockets accept audio.format; WebRTC and SIP negotiate their media format. Voice and format are immutable after startup.

Property Type Required Description
format OpenAI.LiveAudioFormat No Audio encoding and sample rate for audio sent and received over a Live WebSocket connection. WebRTC and SIP negotiate their media format separately.
output OpenAI.LiveInitialSessionAudioOutputParam No The voice used for speech generated by the Live model.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialTextContentPartParam

Assistant text supplied as conversation history when starting a Live session.

Property Type Required Description
type "text" No The text content type. Always text. default: "text".
text string Yes The message text to include in the Live session’s initial conversation history.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInitialUserMessageItemParam

A user message included in the initial text history of a Live session.

All of: OpenAI.LiveInitialItem.

Property Type Required Description
id Any of: string; null No
type "message" No The history item type. Always message. default: "message".
status Any of: OpenAI.LiveInitialMessageStatus; null No
role "user" Yes The author of this history message. Always user.
content Array of OpenAI.LiveInitialInputTextContentPartParam Yes The message content. Supply exactly one text part for the initial Live conversation history. minItems: 1. maxItems: 1.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInlineSkillParam

Property Type Required Description
type "inline" Yes Defines an inline skill for this request. default: "inline".
name string Yes The name of the skill.
description string Yes The description of the skill.
source OpenAI.LiveInlineSkillSourceParam Yes Inline skill payload

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveInlineSkillSourceParam

Inline skill payload

Property Type Required Description
type "base64" Yes The type of the inline skill source. Must be base64. default: "base64".
media_type "application/zip" Yes The media type of the inline skill payload. Must be application/zip. default: "application/zip".
data string Yes Base64-encoded skill zip bundle. minLength: 1. maxLength: 70254592.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveLiveError

Details of an error encountered by the Live session, including the affected parameter or client command when available.

Property Type Required Description
type string Yes The category of error, such as invalid_request_error for an invalid Live client command.
code string Yes A machine-readable code identifying the Live error, such as unknown_parameter.
message string Yes A human-readable explanation of the Live error.
param string No The parameter that caused the error, when applicable, such as session.voice.
client_event_id string No The event_id of the client command that caused the error, when supplied.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveLocalEnvironmentParam

Property Type Required Description
type "local" Yes Use a local computer environment. default: "local".
skills Any of: Array of OpenAI.LiveLocalSkillParam; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveLocalSkillParam

Property Type Required Description
name string Yes The name of the skill.
description string Yes The description of the skill.
path string Yes The path to the directory containing the skill.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveMCPToolChoiceParam

Property Type Required Description
type "mcp" Yes The tool to call. Always mcp. default: "mcp".
server_label string Yes The label of the MCP server to call.
name Any of: string; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveMCPToolInputParam

Property Type Required Description
type "mcp" Yes default: "mcp".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveNamespaceToolInputParam

Property Type Required Description
type "namespace" Yes default: "namespace".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveProgrammaticToolInputParam

Property Type Required Description
type "programmatic_tool_calling" Yes default: "programmatic_tool_calling".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveReasoningEffort

Type: "none" or "minimal" or "low" or "medium" or "high" or "xhigh".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveReasoningSummary

Type: "concise" or "detailed" or "auto".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveResponsesDelegationSettingsInputParam

Model, prompt, and tool settings for tasks delegated by the Live session to a Responses backend.

Property Type Required Description
model string Yes The model used for server-owned Responses delegations.
instructions Any of: string; null No
max_output_tokens Any of: OpenAI.integer; null No
service_tier Any of: OpenAI.LiveResponsesServiceTier; null No
reasoning Any of: OpenAI.LiveDelegationReasoningInputParam; null No
text Any of: OpenAI.LiveDelegationTextInputParam; null No
tools Array of Any of: OpenAI.LiveFunctionToolInputParam; OpenAI.LiveWebSearchToolInputParam; OpenAI.LiveFileSearchToolInputParam; OpenAI.LiveCodeInterpreterToolInputParam; OpenAI.LiveHostedShellToolInputParam; OpenAI.LiveImageGenerationToolInputParam; OpenAI.LiveMCPToolInputParam; OpenAI.LiveCustomToolInputParam; OpenAI.LiveNamespaceToolInputParam; OpenAI.LiveToolSearchToolInputParam; OpenAI.LiveProgrammaticToolInputParam; OpenAI.LiveComputerToolInputParam; OpenAI.LiveApplyPatchToolInputParam No Tools available to the Responses backend while it handles tasks delegated by the Live model.
tool_choice Exactly one of: OpenAI.LiveToolChoiceEnum; OpenAI.LiveFunctionToolChoiceParam; OpenAI.LiveMCPToolChoiceParam; OpenAI.LiveSpecificFileSearchParam; OpenAI.LiveSpecificWebSearchParam; OpenAI.LiveSpecificWebSearchPreviewParam; OpenAI.LiveSpecificImageGenParam; OpenAI.LiveSpecificComputerParam; OpenAI.LiveSpecificCodeInterpreterParam; OpenAI.LiveSpecificProgrammaticToolCallingParam; OpenAI.LiveSpecificFunctionShellParam; OpenAI.LiveSpecificCustomToolParam; OpenAI.LiveSpecificApplyPatchParam; OpenAI.LiveAllowedToolsChoiceParam No Controls which tool the Responses backend uses when handling a task delegated by the Live model.
parallel_tool_calls Any of: boolean; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveResponsesDelegationSettingsUpdateInputParam

Updates to the Responses backend of an existing Live session. Omitted settings retain their current values.

Property Type Required Description
model string No The Responses backend model to use for subsequent delegated requests. Omit to keep the current backend model.
instructions Any of: string; null No
max_output_tokens Any of: OpenAI.integer; null No
service_tier Any of: OpenAI.LiveResponsesServiceTier; null No
reasoning Any of: OpenAI.LiveDelegationReasoningInputParam; null No
text Any of: OpenAI.LiveDelegationTextInputParam; null No
tools Array of Any of: OpenAI.LiveFunctionToolInputParam; OpenAI.LiveWebSearchToolInputParam; OpenAI.LiveFileSearchToolInputParam; OpenAI.LiveCodeInterpreterToolInputParam; OpenAI.LiveHostedShellToolInputParam; OpenAI.LiveImageGenerationToolInputParam; OpenAI.LiveMCPToolInputParam; OpenAI.LiveCustomToolInputParam; OpenAI.LiveNamespaceToolInputParam; OpenAI.LiveToolSearchToolInputParam; OpenAI.LiveProgrammaticToolInputParam; OpenAI.LiveComputerToolInputParam; OpenAI.LiveApplyPatchToolInputParam No Tools available to the Responses backend while it handles tasks delegated by the Live model.
tool_choice Exactly one of: OpenAI.LiveToolChoiceEnum; OpenAI.LiveFunctionToolChoiceParam; OpenAI.LiveMCPToolChoiceParam; OpenAI.LiveSpecificFileSearchParam; OpenAI.LiveSpecificWebSearchParam; OpenAI.LiveSpecificWebSearchPreviewParam; OpenAI.LiveSpecificImageGenParam; OpenAI.LiveSpecificComputerParam; OpenAI.LiveSpecificCodeInterpreterParam; OpenAI.LiveSpecificProgrammaticToolCallingParam; OpenAI.LiveSpecificFunctionShellParam; OpenAI.LiveSpecificCustomToolParam; OpenAI.LiveSpecificApplyPatchParam; OpenAI.LiveAllowedToolsChoiceParam No Controls which tool the Responses backend uses when handling a task delegated by the Live model.
parallel_tool_calls Any of: boolean; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveResponsesDelegationUpdateParam

Update the Responses backend for an existing Live session without changing delegation ownership.

All of: OpenAI.LiveSessionUpdateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "responses" Yes The delegation owner. Always responses for tasks handled by the Responses API.
responses OpenAI.LiveResponsesDelegationSettingsUpdateInputParam No Responses backend settings to update. Omitted settings keep their existing values.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveResponsesServiceTier

Type: "auto" or "default" or "fast_tier_temp_pilot" or "flex" or "priority" or "ultrafast".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveServerEvent

An event sent by the Live API on a primary WebSocket or sideband connection.

Property Type Required Description
type OpenAI.LiveServerEventType Yes

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveServerEventType

Any of: string, "session.started" or "session.updated" or "session.input_audio.muted" or "session.input_audio.unmuted" or "session.output_audio.fixture.playing" or "session.output_audio.fixture.played" or "session.instructions.appended" or "session.thinking.appended" or "session.commentary.appended" or "session.input_audio.append" or "session.output_audio.delta" or "session.input_transcript.delta" or "session.output_transcript.delta" or "session.delegation.created" or "decision.event" or "decision.item.created" or "decision.item.deleted" or "response.event" or "session.usage.updated" or "session.closed" or "error" or "info" or "transport.dtmf.received" or "transport.dtmf.sent" or "transport.ringing" or "transport.answered" or "transport.failed".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionAudioFormatPCMAParam

Raw, mono G.711 A-law audio for a Live WebSocket connection.

All of: OpenAI.LiveAudioFormat.

Property Type Required Description
type "audio/pcma" Yes The audio encoding. Always audio/pcma.
rate OpenAI.integer Yes Audio sample rate in hertz. G.711 audio uses 8000 Hz. minimum: 8000. maximum: 8000.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionAudioFormatPCMParam

Raw, mono 16-bit little-endian PCM audio for a Live WebSocket connection.

All of: OpenAI.LiveAudioFormat.

Property Type Required Description
type "audio/pcm" Yes The audio encoding. Always audio/pcm.
rate 16000 or 24000 Yes Audio sample rate in hertz. Live WebSocket PCM audio supports 16000 or 24000 Hz.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionAudioFormatPCMUParam

Raw, mono G.711 μ-law audio for a Live WebSocket connection.

All of: OpenAI.LiveAudioFormat.

Property Type Required Description
type "audio/pcmu" Yes The audio encoding. Always audio/pcmu.
rate OpenAI.integer Yes Audio sample rate in hertz. G.711 audio uses 8000 Hz. minimum: 8000. maximum: 8000.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionCreateParams

Initial configuration for a Live session, including its model, conversation instructions, audio, and delegated task handling.

Property Type Required Description
model OpenAI.ModelIdsLive Yes
instructions Any of: string; null No
input Array of OpenAI.LiveInitialItem No Ordered text-only history supplied before startup. Supports developer, user, and assistant messages with one text part each; at most 128 messages and a configured rendered-token limit (8,192 by default). maxItems: 128.
audio OpenAI.LiveInitialSessionAudioParam No Startup audio configuration. Only primary WebSockets accept audio.format; WebRTC and SIP negotiate their media format. Voice and format are immutable after startup.
delegation Any of: OpenAI.LiveSessionCreateParamsDelegation; null No
triggers Array of OpenAI.LiveDecisionTriggerInputParam No At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers. maxItems: 2.
store boolean No Whether to store the session for later forking and recording download. Defaults to false for new sessions.
client OpenAI.LiveClientConfigParam No Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionCreateParamsDelegation

Who handles tasks delegated by the Live model. Omitted or null selects your application; use responses or decisions to delegate to the corresponding API.

Variants by type: "client": OpenAI.LiveSessionCreateParamsDelegationLiveClientDelegationParam; "responses": OpenAI.LiveSessionCreateParamsDelegationLiveResponsesDelegationParam; "decisions": OpenAI.LiveSessionCreateParamsDelegationLiveDecisionsDelegationParam.

Property Type Required Description
type OpenAI.LiveSessionCreateParamsDelegationType Yes

OpenAI.LiveSessionCreateParamsDelegationLiveClientDelegationParam

Delegate tasks to your application. The Live session emits delegation events that your backend handles.

All of: OpenAI.LiveSessionCreateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "client" Yes The delegation owner. Always client for tasks handled by your application. default: "client".

OpenAI.LiveSessionCreateParamsDelegationLiveDecisionsDelegationParam

All of: OpenAI.LiveSessionCreateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "decisions" Yes default: "decisions".
decisions OpenAI.LiveDecisionsSettingsInputParam Yes

OpenAI.LiveSessionCreateParamsDelegationLiveResponsesDelegationParam

Delegate tasks to a Responses model managed by the Live session.

All of: OpenAI.LiveSessionCreateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "responses" Yes The delegation owner. Always responses for tasks handled by the Responses API. default: "responses".
responses OpenAI.LiveResponsesDelegationSettingsInputParam Yes Backend model, prompt, and tools used when the Live session delegates a task to Responses.

OpenAI.LiveSessionCreateParamsDelegationType

Any of: string, "client" or "responses" or "decisions".

OpenAI.LiveSessionUpdateParams

Changes to an active Live session's delegation settings and transcript triggers.

Property Type Required Description
delegation Any of: OpenAI.LiveSessionUpdateParamsDelegation; null No
triggers Array of OpenAI.LiveDecisionTriggerInputParam No At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers. maxItems: 2.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSessionUpdateParamsDelegation

Delegation settings to update. The delegation type must match the current session; omitted settings retain their values.

Variants by type: "responses": OpenAI.LiveResponsesDelegationUpdateParam; "decisions": OpenAI.LiveDecisionsDelegationUpdateParam; "client": OpenAI.LiveSessionUpdateParamsDelegationLiveClientDelegationParam.

Property Type Required Description
type OpenAI.LiveSessionUpdateParamsDelegationType Yes

OpenAI.LiveSessionUpdateParamsDelegationLiveClientDelegationParam

Delegate tasks to your application. The Live session emits delegation events that your backend handles.

All of: OpenAI.LiveSessionUpdateParamsDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "client" Yes The delegation owner. Always client for tasks handled by your application. default: "client".

OpenAI.LiveSessionUpdateParamsDelegationType

Any of: string, "client" or "responses" or "decisions".

OpenAI.LiveSessionUsage

Cumulative audio duration for a Live session. Values are totals for the session, not increments to sum across usage events.

Property Type Required Description
seconds OpenAI.numeric Yes The cumulative Live audio duration in seconds. Do not sum this value across usage events.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSkillReferenceParam

Property Type Required Description
type "skill_reference" Yes References a skill created with the /v1/skills endpoint. default: "skill_reference".
skill_id string Yes The ID of the referenced skill. minLength: 1. maxLength: 64.
version Any of: string; null No

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificApplyPatchParam

Forces the model to call the apply_patch tool when executing a tool call.

Property Type Required Description
type "apply_patch" Yes The tool to call. Always apply_patch. default: "apply_patch".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificCodeInterpreterParam

Property Type Required Description
type "code_interpreter" Yes The tool to call. Always code_interpreter. default: "code_interpreter".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificComputerParam

Property Type Required Description
type "computer" Yes The tool to call. Always computer. default: "computer".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificCustomToolParam

Property Type Required Description
type "custom" Yes The tool to call. Always custom. default: "custom".
name string Yes The name of the custom tool to call.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificFileSearchParam

Property Type Required Description
type "file_search" Yes The tool to call. Always file_search. default: "file_search".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificFunctionShellParam

Forces the model to call the shell tool when a tool call is required.

Property Type Required Description
type "shell" Yes The tool to call. Always shell. default: "shell".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificImageGenParam

Property Type Required Description
type "image_generation" Yes The tool to call. Always image_generation. default: "image_generation".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificProgrammaticToolCallingParam

Property Type Required Description
type "programmatic_tool_calling" Yes The tool to call. Always programmatic_tool_calling. default: "programmatic_tool_calling".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificWebSearchParam

Property Type Required Description
type "web_search" Yes The tool to call. Always web_search. default: "web_search".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveSpecificWebSearchPreviewParam

Property Type Required Description
type "web_search_preview" Yes The tool to call. Always web_search_preview. default: "web_search_preview".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveTextVerbosity

Type: "low" or "medium" or "high".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveToolChoiceEnum

Type: "auto" or "none" or "required".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveToolChoiceValueEnum

Type: "none" or "auto" or "required".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveToolSearchToolInputParam

Property Type Required Description
type "tool_search" Yes default: "tool_search".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveTransportCallError

Property Type Required Description
type "call_error" Yes default: "call_error".
code string Yes The call setup failure code.
message string Yes
param string No The parameter related to the error, if any. Empty when no parameter applies.

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.LiveWebSearchToolInputParam

A web search tool available to the Live session’s Responses backend.

Property Type Required Description
type "web_search" Yes The tool type. Always web_search. default: "web_search".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.ModelIdsLive

The Live model. Required in the session configuration for every transport; do not pass it as a URL query parameter.

Any of: string, "gpt-live-1".

Required preview opt-ins: VoiceAgents=V1Preview.

OpenAI.OutputTokensDetailsResource

Property Type Required Description
reasoning_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.

OpenAI.ResponseUsageResource

Property Type Required Description
input_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.
input_tokens_details OpenAI.InputTokensDetailsResource Yes
output_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.
output_tokens_details OpenAI.OutputTokensDetailsResource Yes
total_tokens integer Yes format: "int64". minimum: -2147483648. maximum: 2147483647.

OpenAI.ScoreProbabilityResource

Property Type Required Description
value integer Yes format: "int64".
label string Yes minLength: 0.
probability OpenAI.numeric Yes

OpenAI.integer

Type: integer.

format: "int64".

OpenAI.numeric

Type: number.

format: "double".

VoiceAgentLiveClientDelegation

Delegates work to the application using native client delegation.

All of: VoiceAgentLiveDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
type "client" Yes

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveDelegation

Who handles delegated work for a Live voice agent.

Variants by type: "client": VoiceAgentLiveClientDelegation; "responses": VoiceAgentLiveResponsesDelegation; "foundry_managed_agent": VoiceAgentLiveManagedAgentDelegation.

Property Type Required Description
type VoiceAgentLiveDelegationType Yes The delegation owner.

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveDelegationItem

Live delegation trigger metadata reported to Foundry clients. This object contains metadata, not the task text.

Property Type Required Description
id string Yes The unique ID of the delegation. Use this as delegation_id when replying to client-owned work or correlating backend events.
type "delegation" Yes The object type, always delegation. default: "delegation".
response_id string No The ID of the Responses API response associated with a Responses delegation. Omitted for client delegations.
target VoiceAgentLiveDelegationTarget Yes Where the work is handled. foundry_managed_agent sessions report foundry_managed_agent; GPT Live itself uses client.

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveDelegationTarget

Delegation owner. Foundry agent targets use native client delegation upstream.

Any of: string, "responses" or "client" or "foundry_managed_agent".

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveDelegationType

Supported owners of work delegated by a Live voice agent.

Type: "client" or "responses" or "foundry_managed_agent".

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveManagedAgentDelegation

Delegates to a same-project prompt agent using native client delegation. Requires [thinking] and/or [commentary] output sections.

All of: VoiceAgentLiveDelegation.

Property Type Required Description
type "foundry_managed_agent" Yes Canonical delegation type; legacy managed_prompt_agent values are returned as foundry_managed_agent.
agent_name string Yes Name of a prompt agent in the same project.
agent_version string No Agent version, pinned per connection. Omit to select the active version.

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveResponsesDelegation

Delegates work to a Responses backend using native Responses settings.

All of: VoiceAgentLiveDelegation.

Property Type Required Description
acknowledgement_delay_ms OpenAI.integer No Milliseconds to wait before requesting an automatic spoken acknowledgement for a delegation. The model may still speak before this delay. Any commentary or thinking appended to the session cancels all pending acknowledgements, including those for other delegations. Defaults to 400 for Decisions sessions and 0 (request immediately) for other sessions. Omission in a session update preserves the current value. Updates apply to subsequent delegations. minimum: 0. maximum: 800.
responses OpenAI.LiveResponsesDelegationSettingsInputParam Yes Backend model, prompt, and tools used when the Live session delegates a task to Responses.
type "responses" Yes

Required preview opt-ins: VoiceAgents=V1Preview.

VoiceAgentLiveSessionResource

Live session resource reported to Foundry clients. GPT Live runs native client delegation for foundry_managed_agent sessions; Foundry reports the effective owner.

Property Type Required Description
model OpenAI.ModelIdsLive Yes
instructions Any of: string; null No
input Array of OpenAI.LiveInitialItem No Ordered text-only history supplied before startup. Supports developer, user, and assistant messages with one text part each; at most 128 messages and a configured rendered-token limit (8,192 by default). maxItems: 128.
audio OpenAI.LiveInitialSessionAudioParam No Startup audio configuration. Only primary WebSockets accept audio.format; WebRTC and SIP negotiate their media format. Voice and format are immutable after startup.
triggers Array of OpenAI.LiveDecisionTriggerInputParam No At most one Decisions trigger per transcript event type. Triggers run independently of delegation. On update, omit triggers to keep the current configuration, send a list to replace it, or send an empty list to disable triggers. maxItems: 2.
store boolean No Whether to store the session for later forking and recording download. Defaults to false for new sessions.
client OpenAI.LiveClientConfigParam No Startup-only capabilities for an untrusted frontend attached to a unified WebRTC session. Trusted sideband connections are unaffected.
id string Yes The unique ID of the Live session. Use this ID for sideband connections, forking, and recording download.
expires_at OpenAI.integer Yes The Unix timestamp, in seconds, at which the Live session expires.
status "active" Yes The status of the session snapshot. Always active, including the final snapshot in session.closed; use the event type to determine that the session has closed. default: "active".
delegation Any of: VoiceAgentLiveDelegation; null No Delegation owner. foundry_managed_agent sessions report the pinned target, including the resolved agent_version.

Required preview opt-ins: VoiceAgents=V1Preview.