Create client secret
Create a Realtime client secret with an associated session configuration.
ParametersExpand Collapse
Configuration for the client secret expiration. Expiration refers to the time after which
a client secret will no longer be valid for creating sessions. The session itself may
continue after that time once started. A secret can be used to create multiple sessions
until it expires.
Configuration for the client secret expiration. Expiration refers to the time after which a client secret will no longer be valid for creating sessions. The session itself may continue after that time once started. A secret can be used to create multiple sessions until it expires.
The anchor point for the client secret expiration, meaning that seconds will be added to the created_at time of the client secret to produce an expiration timestamp. Only created_at is currently supported.
The number of seconds from the anchor point to the expiration. Select a value between 10 and 7200 (2 hours). This default to 600 seconds (10 minutes) if not specified.
Session configuration to use for the client secret. Choose either a realtime
session or a transcription session.
Session configuration to use for the client secret. Choose either a realtime session or a transcription session.
class RealtimeSessionCreateRequest: …Realtime session object configuration.
Realtime session object configuration.
The type of session to create. Always realtime for the Realtime API.
audio: Optional[RealtimeAudioConfig]Configuration for input and output audio.
Configuration for input and output audio.
input: Optional[RealtimeAudioConfigInput]
format: Optional[RealtimeAudioFormats]The format of the input audio.
The format of the input audio.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
noise_reduction: Optional[NoiseReduction]Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
type: Optional[NoiseReductionType]Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
transcription: Optional[AudioTranscription]Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
The language of the input audio. Supplying the input language in
ISO-639-1 (e.g. en) format
will improve accuracy and latency.
model: Optional[Union[str, Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more], null]]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
An optional text to guide the model's style or continue a previous audio
segment.
For whisper-1, the prompt is a list of keywords.
For gpt-4o-transcribe models (excluding gpt-4o-transcribe-diarize), the prompt is a free text string, for example "expect words related to technology".
turn_detection: Optional[RealtimeAudioInputTurnDetection]Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
class ServerVad: …Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Type of turn detection, server_vad to turn on simple Server VAD.
Whether or not to automatically generate a response when a VAD stop event occurs. If interrupt_response is set to false this may fail to create a response if the model is already responding.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Optional timeout after which a model response will be triggered automatically. This is useful for situations in which a long pause from the user is unexpected, such as a phone call. The model will effectively prompt the user to continue the conversation based on the current context.
The timeout value will be applied after the last model response's audio has finished playing,
i.e. it's set to the response.done time plus audio playback duration.
An input_audio_buffer.timeout_triggered event (plus events
associated with the Response) will be emitted when the timeout is reached.
Idle timeout is currently only supported for server_vad mode.
Whether or not to automatically interrupt (cancel) any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs. If true then the response will be cancelled, otherwise it will continue until complete.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Used only for server_vad mode. Amount of audio to include before the VAD detected speech (in
milliseconds). Defaults to 300ms.
Used only for server_vad mode. Duration of silence to detect speech stop (in milliseconds). Defaults
to 500ms. With shorter values the model will respond more quickly,
but may jump in on short pauses from the user.
Used only for server_vad mode. Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A
higher threshold will require louder audio to activate the model, and
thus might perform better in noisy environments.
class SemanticVad: …Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Type of turn detection, semantic_vad to turn on Semantic VAD.
Whether or not to automatically generate a response when a VAD stop event occurs.
eagerness: Optional[Literal["low", "medium", "high", "auto"]]Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Whether or not to automatically interrupt any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs.
output: Optional[RealtimeAudioConfigOutput]
format: Optional[RealtimeAudioFormats]The format of the output audio.
The format of the output audio.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
The speed of the model's spoken response as a multiple of the original speed. 1.0 is the default speed. 0.25 is the minimum speed. 1.5 is the maximum speed. This value can only be changed in between model turns, not while a response is in progress.
This parameter is a post-processing adjustment to the audio after it is generated, it's also possible to prompt the model to speak faster or slower.
voice: Optional[Union[str, Literal["alloy", "ash", "ballad", 7 more], null]]The voice the model uses to respond. Supported built-in voices are alloy, ash, ballad, coral, echo, sage, shimmer, verse, marin, and cedar. Voice cannot be changed during the session once the model has responded with audio at least once. We recommend marin and cedar for best quality.
The voice the model uses to respond. Supported built-in voices are alloy, ash, ballad, coral, echo, sage, shimmer, verse, marin, and cedar. Voice cannot be changed during the session once the model has responded with audio at least once. We recommend marin and cedar for best quality.
Literal["alloy", "ash", "ballad", 7 more]The voice the model uses to respond. Supported built-in voices are alloy, ash, ballad, coral, echo, sage, shimmer, verse, marin, and cedar. Voice cannot be changed during the session once the model has responded with audio at least once. We recommend marin and cedar for best quality.
The voice the model uses to respond. Supported built-in voices are alloy, ash, ballad, coral, echo, sage, shimmer, verse, marin, and cedar. Voice cannot be changed during the session once the model has responded with audio at least once. We recommend marin and cedar for best quality.
Additional fields to include in server outputs.
item.input_audio_transcription.logprobs: Include logprobs for input audio transcription.
The default system instructions (i.e. system message) prepended to model calls. This field allows the client to guide the model on desired responses. The model can be instructed on response content and format, (e.g. "be extremely succinct", "act friendly", "here are examples of good responses") and on audio behavior (e.g. "talk quickly", "inject emotion into your voice", "laugh frequently"). The instructions are not guaranteed to be followed by the model, but they provide guidance to the model on the desired behavior.
Note that the server sets default instructions which will be used if this field is not set and are visible in the session.created event at the start of the session.
max_output_tokens: Optional[Union[int, Literal["inf"], null]]Maximum number of output tokens for a single assistant response,
inclusive of tool calls. Provide an integer between 1 and 4096 to
limit output tokens, or inf for the maximum available tokens for a
given model. Defaults to inf.
Maximum number of output tokens for a single assistant response,
inclusive of tool calls. Provide an integer between 1 and 4096 to
limit output tokens, or inf for the maximum available tokens for a
given model. Defaults to inf.
model: Optional[Union[str, Literal["gpt-realtime", "gpt-realtime-2025-08-28", "gpt-4o-realtime-preview", 11 more], null]]The Realtime model used for this session.
The Realtime model used for this session.
Literal["gpt-realtime", "gpt-realtime-2025-08-28", "gpt-4o-realtime-preview", 11 more]The Realtime model used for this session.
The Realtime model used for this session.
output_modalities: Optional[List[Literal["text", "audio"]]]The set of modalities the model can respond with. It defaults to ["audio"], indicating
that the model will respond with audio plus a transcript. ["text"] can be used to make
the model respond with text only. It is not possible to request both text and audio at the same time.
The set of modalities the model can respond with. It defaults to ["audio"], indicating
that the model will respond with audio plus a transcript. ["text"] can be used to make
the model respond with text only. It is not possible to request both text and audio at the same time.
prompt: Optional[ResponsePrompt]Reference to a prompt template and its variables.
Learn more.
Reference to a prompt template and its variables. Learn more.
The unique identifier of the prompt template to use.
variables: Optional[Dict[str, Variables]]Optional map of values to substitute in for variables in your
prompt. The substitution values can either be strings, or other
Response input types like images or files.
Optional map of values to substitute in for variables in your prompt. The substitution values can either be strings, or other Response input types like images or files.
class ResponseInputText: …A text input to the model.
A text input to the model.
The text input to the model.
The type of the input item. Always input_text.
class ResponseInputImage: …An image input to the model. Learn about image inputs.
An image input to the model. Learn about image inputs.
detail: Literal["low", "high", "auto"]The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.
The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.
The type of the input item. Always input_image.
The ID of the file to be sent to the model.
The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
class ResponseInputFile: …A file input to the model.
A file input to the model.
The type of the input item. Always input_file.
The content of the file to be sent to the model.
The ID of the file to be sent to the model.
The URL of the file to be sent to the model.
The name of the file to be sent to the model.
Optional version of the prompt template.
tool_choice: Optional[RealtimeToolChoiceConfig]How the model chooses tools. Provide one of the string modes or force a specific
function/MCP tool.
How the model chooses tools. Provide one of the string modes or force a specific function/MCP tool.
Literal["none", "auto", "required"]
class ToolChoiceFunction: …Use this option to force the model to call a specific function.
Use this option to force the model to call a specific function.
The name of the function to call.
For function calling, the type is always function.
class ToolChoiceMcp: …Use this option to force the model to call a specific tool on a remote MCP server.
Use this option to force the model to call a specific tool on a remote MCP server.
The label of the MCP server to use.
For MCP tools, the type is always mcp.
The name of the tool to call on the server.
tools: Optional[RealtimeToolsConfig]Tools available to the model.
Tools available to the model.
class RealtimeFunctionTool: …
The description of the function, including guidance on when and how to call it, and guidance about what to tell the user when calling (if anything).
The name of the function.
Parameters of the function in JSON Schema.
The type of the tool, i.e. function.
class Mcp: …Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
Give the model access to additional tools via remote Model Context Protocol (MCP) servers. Learn more about MCP.
A label for this MCP server, used to identify it in tool calls.
The type of the MCP tool. Always mcp.
allowed_tools: Optional[McpAllowedTools]List of allowed tool names or a filter object.
List of allowed tool names or a filter object.
A string array of allowed tool names
class McpAllowedToolsMcpToolFilter: …A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
An OAuth access token that can be used with a remote MCP server, either with a custom MCP server URL or a service connector. Your application must handle the OAuth authorization flow and provide the token here.
connector_id: Optional[Literal["connector_dropbox", "connector_gmail", "connector_googlecalendar", 5 more]]Identifier for service connectors, like those available in ChatGPT. One of
server_url or connector_id must be provided. Learn more about service
connectors here.
Currently supported connector_id values are:
- Dropbox:
connector_dropbox
- Gmail:
connector_gmail
- Google Calendar:
connector_googlecalendar
- Google Drive:
connector_googledrive
- Microsoft Teams:
connector_microsoftteams
- Outlook Calendar:
connector_outlookcalendar
- Outlook Email:
connector_outlookemail
- SharePoint:
connector_sharepoint
Identifier for service connectors, like those available in ChatGPT. One of
server_url or connector_id must be provided. Learn more about service
connectors here.
Currently supported connector_id values are:
- Dropbox:
connector_dropbox - Gmail:
connector_gmail - Google Calendar:
connector_googlecalendar - Google Drive:
connector_googledrive - Microsoft Teams:
connector_microsoftteams - Outlook Calendar:
connector_outlookcalendar - Outlook Email:
connector_outlookemail - SharePoint:
connector_sharepoint
Optional HTTP headers to send to the MCP server. Use for authentication or other purposes.
require_approval: Optional[McpRequireApproval]Specify which of the MCP server's tools require approval.
Specify which of the MCP server's tools require approval.
class McpRequireApprovalMcpToolApprovalFilter: …Specify which of the MCP server's tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
Specify which of the MCP server's tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: Optional[McpRequireApprovalMcpToolApprovalFilterAlways]A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
never: Optional[McpRequireApprovalMcpToolApprovalFilterNever]A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
Literal["always", "never"]Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
Optional description of the MCP server, used to provide more context.
The URL for the MCP server. One of server_url or connector_id must be
provided.
tracing: Optional[RealtimeTracingConfig]Realtime API can write session traces to the Traces Dashboard. Set to null to disable tracing. Once
tracing is enabled for a session, the configuration cannot be modified.
auto will create a trace for the session with default values for the
workflow name, group id, and metadata.
Realtime API can write session traces to the Traces Dashboard. Set to null to disable tracing. Once tracing is enabled for a session, the configuration cannot be modified.
auto will create a trace for the session with default values for the
workflow name, group id, and metadata.
Enables tracing and sets default values for tracing configuration options. Always auto.
class TracingConfiguration: …Granular configuration for tracing.
Granular configuration for tracing.
The group id to attach to this trace to enable filtering and grouping in the Traces Dashboard.
The arbitrary metadata to attach to this trace to enable filtering in the Traces Dashboard.
The name of the workflow to attach to this trace. This is used to name the trace in the Traces Dashboard.
truncation: Optional[RealtimeTruncation]When the number of tokens in a conversation exceeds the model's input token limit, the conversation be truncated, meaning messages (starting from the oldest) will not be included in the model's context. A 32k context model with 4,096 max output tokens can only include 28,224 tokens in the context before truncation occurs.
Clients can configure truncation behavior to truncate with a lower max token limit, which is an effective way to control token usage and cost.
Truncation will reduce the number of cached tokens on the next turn (busting the cache), since messages are dropped from the beginning of the context. However, clients can also configure truncation to retain messages up to a fraction of the maximum context size, which will reduce the need for future truncations and thus improve the cache rate.
Truncation can be disabled entirely, which means the server will never truncate but would instead return an error if the conversation exceeds the model's input token limit.
When the number of tokens in a conversation exceeds the model's input token limit, the conversation be truncated, meaning messages (starting from the oldest) will not be included in the model's context. A 32k context model with 4,096 max output tokens can only include 28,224 tokens in the context before truncation occurs.
Clients can configure truncation behavior to truncate with a lower max token limit, which is an effective way to control token usage and cost.
Truncation will reduce the number of cached tokens on the next turn (busting the cache), since messages are dropped from the beginning of the context. However, clients can also configure truncation to retain messages up to a fraction of the maximum context size, which will reduce the need for future truncations and thus improve the cache rate.
Truncation can be disabled entirely, which means the server will never truncate but would instead return an error if the conversation exceeds the model's input token limit.
Literal["auto", "disabled"]The truncation strategy to use for the session. auto is the default truncation strategy. disabled will disable truncation and emit errors when the conversation exceeds the input token limit.
The truncation strategy to use for the session. auto is the default truncation strategy. disabled will disable truncation and emit errors when the conversation exceeds the input token limit.
class RealtimeTruncationRetentionRatio: …Retain a fraction of the conversation tokens when the conversation exceeds the input token limit. This allows you to amortize truncations across multiple turns, which can help improve cached token usage.
Retain a fraction of the conversation tokens when the conversation exceeds the input token limit. This allows you to amortize truncations across multiple turns, which can help improve cached token usage.
Fraction of post-instruction conversation tokens to retain (0.0 - 1.0) when the conversation exceeds the input token limit. Setting this to 0.8 means that messages will be dropped until 80% of the maximum allowed tokens are used. This helps reduce the frequency of truncations and improve cache rates.
Use retention ratio truncation.
token_limits: Optional[TokenLimits]Optional custom token limits for this truncation strategy. If not provided, the model's default token limits will be used.
Optional custom token limits for this truncation strategy. If not provided, the model's default token limits will be used.
Maximum tokens allowed in the conversation after instructions (which including tool definitions). For example, setting this to 5,000 would mean that truncation would occur when the conversation exceeds 5,000 tokens after instructions. This cannot be higher than the model's context window size minus the maximum output tokens.
class RealtimeTranscriptionSessionCreateRequest: …Realtime transcription session object configuration.
Realtime transcription session object configuration.
The type of session to create. Always transcription for transcription sessions.
audio: Optional[RealtimeTranscriptionSessionAudio]Configuration for input and output audio.
Configuration for input and output audio.
input: Optional[RealtimeTranscriptionSessionAudioInput]
format: Optional[RealtimeAudioFormats]The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
noise_reduction: Optional[NoiseReduction]Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
type: Optional[NoiseReductionType]Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
transcription: Optional[AudioTranscription]Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
The language of the input audio. Supplying the input language in
ISO-639-1 (e.g. en) format
will improve accuracy and latency.
model: Optional[Union[str, Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more], null]]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
An optional text to guide the model's style or continue a previous audio
segment.
For whisper-1, the prompt is a list of keywords.
For gpt-4o-transcribe models (excluding gpt-4o-transcribe-diarize), the prompt is a free text string, for example "expect words related to technology".
turn_detection: Optional[RealtimeTranscriptionSessionAudioInputTurnDetection]Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
class ServerVad: …Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Type of turn detection, server_vad to turn on simple Server VAD.
Whether or not to automatically generate a response when a VAD stop event occurs. If interrupt_response is set to false this may fail to create a response if the model is already responding.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Optional timeout after which a model response will be triggered automatically. This is useful for situations in which a long pause from the user is unexpected, such as a phone call. The model will effectively prompt the user to continue the conversation based on the current context.
The timeout value will be applied after the last model response's audio has finished playing,
i.e. it's set to the response.done time plus audio playback duration.
An input_audio_buffer.timeout_triggered event (plus events
associated with the Response) will be emitted when the timeout is reached.
Idle timeout is currently only supported for server_vad mode.
Whether or not to automatically interrupt (cancel) any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs. If true then the response will be cancelled, otherwise it will continue until complete.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Used only for server_vad mode. Amount of audio to include before the VAD detected speech (in
milliseconds). Defaults to 300ms.
Used only for server_vad mode. Duration of silence to detect speech stop (in milliseconds). Defaults
to 500ms. With shorter values the model will respond more quickly,
but may jump in on short pauses from the user.
Used only for server_vad mode. Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A
higher threshold will require louder audio to activate the model, and
thus might perform better in noisy environments.
class SemanticVad: …Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Type of turn detection, semantic_vad to turn on Semantic VAD.
Whether or not to automatically generate a response when a VAD stop event occurs.
eagerness: Optional[Literal["low", "medium", "high", "auto"]]Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Whether or not to automatically interrupt any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs.
Additional fields to include in server outputs.
item.input_audio_transcription.logprobs: Include logprobs for input audio transcription.
ReturnsExpand Collapse
class ClientSecretCreateResponse: …Response from creating a session and client secret for the Realtime API.
Response from creating a session and client secret for the Realtime API.
Expiration timestamp for the client secret, in seconds since epoch.
session: SessionThe session configuration for either a realtime or transcription session.
The session configuration for either a realtime or transcription session.
class RealtimeSessionCreateResponse: …A new Realtime session configuration, with an ephemeral key. Default TTL
for keys is one minute.
A new Realtime session configuration, with an ephemeral key. Default TTL for keys is one minute.
Ephemeral key returned by the API.
Ephemeral key returned by the API.
Timestamp for when the token expires. Currently, all tokens expire after one minute.
Ephemeral key usable in client environments to authenticate connections to the Realtime API. Use this in client-side environments rather than a standard API token, which should only be used server-side.
The type of session to create. Always realtime for the Realtime API.
audio: Optional[Audio]Configuration for input and output audio.
Configuration for input and output audio.
input: Optional[AudioInput]
format: Optional[RealtimeAudioFormats]The format of the input audio.
The format of the input audio.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
noise_reduction: Optional[AudioInputNoiseReduction]Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
Configuration for input audio noise reduction. This can be set to null to turn off.
Noise reduction filters audio added to the input audio buffer before it is sent to VAD and the model.
Filtering the audio can improve VAD and turn detection accuracy (reducing false positives) and model performance by improving perception of the input audio.
type: Optional[NoiseReductionType]Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
transcription: Optional[AudioTranscription]Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
Configuration for input audio transcription, defaults to off and can be set to null to turn off once on. Input audio transcription is not native to the model, since the model consumes audio directly. Transcription runs asynchronously through the /audio/transcriptions endpoint and should be treated as guidance of input audio content rather than precisely what the model heard. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.
The language of the input audio. Supplying the input language in
ISO-639-1 (e.g. en) format
will improve accuracy and latency.
model: Optional[Union[str, Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more], null]]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
An optional text to guide the model's style or continue a previous audio
segment.
For whisper-1, the prompt is a list of keywords.
For gpt-4o-transcribe models (excluding gpt-4o-transcribe-diarize), the prompt is a free text string, for example "expect words related to technology".
turn_detection: Optional[AudioInputTurnDetection]Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
Configuration for turn detection, ether Server VAD or Semantic VAD. This can be set to null to turn off, in which case the client must manually trigger model response.
Server VAD means that the model will detect the start and end of speech based on audio volume and respond at the end of user speech.
Semantic VAD is more advanced and uses a turn detection model (in conjunction with VAD) to semantically estimate whether the user has finished speaking, then dynamically sets a timeout based on this probability. For example, if user audio trails off with "uhhm", the model will score a low probability of turn end and wait longer for the user to continue speaking. This can be useful for more natural conversations, but may have a higher latency.
class AudioInputTurnDetectionServerVad: …Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Server-side voice activity detection (VAD) which flips on when user speech is detected and off after a period of silence.
Type of turn detection, server_vad to turn on simple Server VAD.
Whether or not to automatically generate a response when a VAD stop event occurs. If interrupt_response is set to false this may fail to create a response if the model is already responding.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Optional timeout after which a model response will be triggered automatically. This is useful for situations in which a long pause from the user is unexpected, such as a phone call. The model will effectively prompt the user to continue the conversation based on the current context.
The timeout value will be applied after the last model response's audio has finished playing,
i.e. it's set to the response.done time plus audio playback duration.
An input_audio_buffer.timeout_triggered event (plus events
associated with the Response) will be emitted when the timeout is reached.
Idle timeout is currently only supported for server_vad mode.
Whether or not to automatically interrupt (cancel) any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs. If true then the response will be cancelled, otherwise it will continue until complete.
If both create_response and interrupt_response are set to false, the model will never respond automatically but VAD events will still be emitted.
Used only for server_vad mode. Amount of audio to include before the VAD detected speech (in
milliseconds). Defaults to 300ms.
Used only for server_vad mode. Duration of silence to detect speech stop (in milliseconds). Defaults
to 500ms. With shorter values the model will respond more quickly,
but may jump in on short pauses from the user.
Used only for server_vad mode. Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A
higher threshold will require louder audio to activate the model, and
thus might perform better in noisy environments.
class AudioInputTurnDetectionSemanticVad: …Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Server-side semantic turn detection which uses a model to determine when the user has finished speaking.
Type of turn detection, semantic_vad to turn on Semantic VAD.
Whether or not to automatically generate a response when a VAD stop event occurs.
eagerness: Optional[Literal["low", "medium", "high", "auto"]]Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Used only for semantic_vad mode. The eagerness of the model to respond. low will wait longer for the user to continue speaking, high will respond more quickly. auto is the default and is equivalent to medium. low, medium, and high have max timeouts of 8s, 4s, and 2s respectively.
Whether or not to automatically interrupt any ongoing response with output to the default
conversation (i.e. conversation of auto) when a VAD start event occurs.
output: Optional[AudioOutput]
format: Optional[RealtimeAudioFormats]The format of the output audio.
The format of the output audio.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
The speed of the model's spoken response as a multiple of the original speed. 1.0 is the default speed. 0.25 is the minimum speed. 1.5 is the maximum speed. This value can only be changed in between model turns, not while a response is in progress.
This parameter is a post-processing adjustment to the audio after it is generated, it's also possible to prompt the model to speak faster or slower.
voice: Optional[Union[str, Literal["alloy", "ash", "ballad", 7 more], null]]The voice the model uses to respond. Voice cannot be changed during the
session once the model has responded with audio at least once. Current
voice options are alloy, ash, ballad, coral, echo, sage,
shimmer, verse, marin, and cedar. We recommend marin and cedar for
best quality.
The voice the model uses to respond. Voice cannot be changed during the
session once the model has responded with audio at least once. Current
voice options are alloy, ash, ballad, coral, echo, sage,
shimmer, verse, marin, and cedar. We recommend marin and cedar for
best quality.
Literal["alloy", "ash", "ballad", 7 more]The voice the model uses to respond. Voice cannot be changed during the
session once the model has responded with audio at least once. Current
voice options are alloy, ash, ballad, coral, echo, sage,
shimmer, verse, marin, and cedar. We recommend marin and cedar for
best quality.
The voice the model uses to respond. Voice cannot be changed during the
session once the model has responded with audio at least once. Current
voice options are alloy, ash, ballad, coral, echo, sage,
shimmer, verse, marin, and cedar. We recommend marin and cedar for
best quality.
Additional fields to include in server outputs.
item.input_audio_transcription.logprobs: Include logprobs for input audio transcription.
The default system instructions (i.e. system message) prepended to model calls. This field allows the client to guide the model on desired responses. The model can be instructed on response content and format, (e.g. "be extremely succinct", "act friendly", "here are examples of good responses") and on audio behavior (e.g. "talk quickly", "inject emotion into your voice", "laugh frequently"). The instructions are not guaranteed to be followed by the model, but they provide guidance to the model on the desired behavior.
Note that the server sets default instructions which will be used if this field is not set and are visible in the session.created event at the start of the session.
max_output_tokens: Optional[Union[int, Literal["inf"], null]]Maximum number of output tokens for a single assistant response,
inclusive of tool calls. Provide an integer between 1 and 4096 to
limit output tokens, or inf for the maximum available tokens for a
given model. Defaults to inf.
Maximum number of output tokens for a single assistant response,
inclusive of tool calls. Provide an integer between 1 and 4096 to
limit output tokens, or inf for the maximum available tokens for a
given model. Defaults to inf.
model: Optional[Union[str, Literal["gpt-realtime", "gpt-realtime-2025-08-28", "gpt-4o-realtime-preview", 11 more], null]]The Realtime model used for this session.
The Realtime model used for this session.
Literal["gpt-realtime", "gpt-realtime-2025-08-28", "gpt-4o-realtime-preview", 11 more]The Realtime model used for this session.
The Realtime model used for this session.
output_modalities: Optional[List[Literal["text", "audio"]]]The set of modalities the model can respond with. It defaults to ["audio"], indicating
that the model will respond with audio plus a transcript. ["text"] can be used to make
the model respond with text only. It is not possible to request both text and audio at the same time.
The set of modalities the model can respond with. It defaults to ["audio"], indicating
that the model will respond with audio plus a transcript. ["text"] can be used to make
the model respond with text only. It is not possible to request both text and audio at the same time.
prompt: Optional[ResponsePrompt]Reference to a prompt template and its variables.
Learn more.
Reference to a prompt template and its variables. Learn more.
The unique identifier of the prompt template to use.
variables: Optional[Dict[str, Variables]]Optional map of values to substitute in for variables in your
prompt. The substitution values can either be strings, or other
Response input types like images or files.
Optional map of values to substitute in for variables in your prompt. The substitution values can either be strings, or other Response input types like images or files.
class ResponseInputText: …A text input to the model.
A text input to the model.
The text input to the model.
The type of the input item. Always input_text.
class ResponseInputImage: …An image input to the model. Learn about image inputs.
An image input to the model. Learn about image inputs.
detail: Literal["low", "high", "auto"]The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.
The detail level of the image to be sent to the model. One of high, low, or auto. Defaults to auto.
The type of the input item. Always input_image.
The ID of the file to be sent to the model.
The URL of the image to be sent to the model. A fully qualified URL or base64 encoded image in a data URL.
class ResponseInputFile: …A file input to the model.
A file input to the model.
The type of the input item. Always input_file.
The content of the file to be sent to the model.
The ID of the file to be sent to the model.
The URL of the file to be sent to the model.
The name of the file to be sent to the model.
Optional version of the prompt template.
tool_choice: Optional[ToolChoice]How the model chooses tools. Provide one of the string modes or force a specific
function/MCP tool.
How the model chooses tools. Provide one of the string modes or force a specific function/MCP tool.
Literal["none", "auto", "required"]
class ToolChoiceFunction: …Use this option to force the model to call a specific function.
Use this option to force the model to call a specific function.
The name of the function to call.
For function calling, the type is always function.
class ToolChoiceMcp: …Use this option to force the model to call a specific tool on a remote MCP server.
Use this option to force the model to call a specific tool on a remote MCP server.
The label of the MCP server to use.
For MCP tools, the type is always mcp.
The name of the tool to call on the server.
tools: Optional[List[Tool]]Tools available to the model.
Tools available to the model.
class RealtimeFunctionTool: …
The description of the function, including guidance on when and how to call it, and guidance about what to tell the user when calling (if anything).
The name of the function.
Parameters of the function in JSON Schema.
The type of the tool, i.e. function.
class ToolMcpTool: …Give the model access to additional tools via remote Model Context Protocol
(MCP) servers. Learn more about MCP.
Give the model access to additional tools via remote Model Context Protocol (MCP) servers. Learn more about MCP.
A label for this MCP server, used to identify it in tool calls.
The type of the MCP tool. Always mcp.
allowed_tools: Optional[ToolMcpToolAllowedTools]List of allowed tool names or a filter object.
List of allowed tool names or a filter object.
A string array of allowed tool names
class ToolMcpToolAllowedToolsMcpToolFilter: …A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
An OAuth access token that can be used with a remote MCP server, either with a custom MCP server URL or a service connector. Your application must handle the OAuth authorization flow and provide the token here.
connector_id: Optional[Literal["connector_dropbox", "connector_gmail", "connector_googlecalendar", 5 more]]Identifier for service connectors, like those available in ChatGPT. One of
server_url or connector_id must be provided. Learn more about service
connectors here.
Currently supported connector_id values are:
- Dropbox:
connector_dropbox
- Gmail:
connector_gmail
- Google Calendar:
connector_googlecalendar
- Google Drive:
connector_googledrive
- Microsoft Teams:
connector_microsoftteams
- Outlook Calendar:
connector_outlookcalendar
- Outlook Email:
connector_outlookemail
- SharePoint:
connector_sharepoint
Identifier for service connectors, like those available in ChatGPT. One of
server_url or connector_id must be provided. Learn more about service
connectors here.
Currently supported connector_id values are:
- Dropbox:
connector_dropbox - Gmail:
connector_gmail - Google Calendar:
connector_googlecalendar - Google Drive:
connector_googledrive - Microsoft Teams:
connector_microsoftteams - Outlook Calendar:
connector_outlookcalendar - Outlook Email:
connector_outlookemail - SharePoint:
connector_sharepoint
Optional HTTP headers to send to the MCP server. Use for authentication or other purposes.
require_approval: Optional[ToolMcpToolRequireApproval]Specify which of the MCP server's tools require approval.
Specify which of the MCP server's tools require approval.
class ToolMcpToolRequireApprovalMcpToolApprovalFilter: …Specify which of the MCP server's tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
Specify which of the MCP server's tools require approval. Can be
always, never, or a filter object associated with tools
that require approval.
always: Optional[ToolMcpToolRequireApprovalMcpToolApprovalFilterAlways]A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
never: Optional[ToolMcpToolRequireApprovalMcpToolApprovalFilterNever]A filter object to specify which tools are allowed.
A filter object to specify which tools are allowed.
Indicates whether or not a tool modifies data or is read-only. If an
MCP server is annotated with readOnlyHint,
it will match this filter.
List of allowed tool names.
Literal["always", "never"]Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
Specify a single approval policy for all tools. One of always or
never. When set to always, all tools will require approval. When
set to never, all tools will not require approval.
Optional description of the MCP server, used to provide more context.
The URL for the MCP server. One of server_url or connector_id must be
provided.
tracing: Optional[Tracing]Realtime API can write session traces to the Traces Dashboard. Set to null to disable tracing. Once
tracing is enabled for a session, the configuration cannot be modified.
auto will create a trace for the session with default values for the
workflow name, group id, and metadata.
Realtime API can write session traces to the Traces Dashboard. Set to null to disable tracing. Once tracing is enabled for a session, the configuration cannot be modified.
auto will create a trace for the session with default values for the
workflow name, group id, and metadata.
Enables tracing and sets default values for tracing configuration options. Always auto.
class TracingTracingConfiguration: …Granular configuration for tracing.
Granular configuration for tracing.
The group id to attach to this trace to enable filtering and grouping in the Traces Dashboard.
The arbitrary metadata to attach to this trace to enable filtering in the Traces Dashboard.
The name of the workflow to attach to this trace. This is used to name the trace in the Traces Dashboard.
truncation: Optional[RealtimeTruncation]When the number of tokens in a conversation exceeds the model's input token limit, the conversation be truncated, meaning messages (starting from the oldest) will not be included in the model's context. A 32k context model with 4,096 max output tokens can only include 28,224 tokens in the context before truncation occurs.
Clients can configure truncation behavior to truncate with a lower max token limit, which is an effective way to control token usage and cost.
Truncation will reduce the number of cached tokens on the next turn (busting the cache), since messages are dropped from the beginning of the context. However, clients can also configure truncation to retain messages up to a fraction of the maximum context size, which will reduce the need for future truncations and thus improve the cache rate.
Truncation can be disabled entirely, which means the server will never truncate but would instead return an error if the conversation exceeds the model's input token limit.
When the number of tokens in a conversation exceeds the model's input token limit, the conversation be truncated, meaning messages (starting from the oldest) will not be included in the model's context. A 32k context model with 4,096 max output tokens can only include 28,224 tokens in the context before truncation occurs.
Clients can configure truncation behavior to truncate with a lower max token limit, which is an effective way to control token usage and cost.
Truncation will reduce the number of cached tokens on the next turn (busting the cache), since messages are dropped from the beginning of the context. However, clients can also configure truncation to retain messages up to a fraction of the maximum context size, which will reduce the need for future truncations and thus improve the cache rate.
Truncation can be disabled entirely, which means the server will never truncate but would instead return an error if the conversation exceeds the model's input token limit.
Literal["auto", "disabled"]The truncation strategy to use for the session. auto is the default truncation strategy. disabled will disable truncation and emit errors when the conversation exceeds the input token limit.
The truncation strategy to use for the session. auto is the default truncation strategy. disabled will disable truncation and emit errors when the conversation exceeds the input token limit.
class RealtimeTruncationRetentionRatio: …Retain a fraction of the conversation tokens when the conversation exceeds the input token limit. This allows you to amortize truncations across multiple turns, which can help improve cached token usage.
Retain a fraction of the conversation tokens when the conversation exceeds the input token limit. This allows you to amortize truncations across multiple turns, which can help improve cached token usage.
Fraction of post-instruction conversation tokens to retain (0.0 - 1.0) when the conversation exceeds the input token limit. Setting this to 0.8 means that messages will be dropped until 80% of the maximum allowed tokens are used. This helps reduce the frequency of truncations and improve cache rates.
Use retention ratio truncation.
token_limits: Optional[TokenLimits]Optional custom token limits for this truncation strategy. If not provided, the model's default token limits will be used.
Optional custom token limits for this truncation strategy. If not provided, the model's default token limits will be used.
Maximum tokens allowed in the conversation after instructions (which including tool definitions). For example, setting this to 5,000 would mean that truncation would occur when the conversation exceeds 5,000 tokens after instructions. This cannot be higher than the model's context window size minus the maximum output tokens.
class RealtimeTranscriptionSessionCreateResponse: …A Realtime transcription session configuration object.
A Realtime transcription session configuration object.
Unique identifier for the session that looks like sess_1234567890abcdef.
The object type. Always realtime.transcription_session.
The type of session. Always transcription for transcription sessions.
audio: Optional[Audio]Configuration for input audio for the session.
Configuration for input audio for the session.
input: Optional[AudioInput]
format: Optional[RealtimeAudioFormats]The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
class AudioPCM: …The PCM audio format. Only a 24kHz sample rate is supported.
The PCM audio format. Only a 24kHz sample rate is supported.
The sample rate of the audio. Always 24000.
The audio format. Always audio/pcm.
class AudioPCMU: …The G.711 μ-law format.
The G.711 μ-law format.
The audio format. Always audio/pcmu.
class AudioPCMA: …The G.711 A-law format.
The G.711 A-law format.
The audio format. Always audio/pcma.
noise_reduction: Optional[AudioInputNoiseReduction]Configuration for input audio noise reduction.
Configuration for input audio noise reduction.
type: Optional[NoiseReductionType]Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
Type of noise reduction. near_field is for close-talking microphones such as headphones, far_field is for far-field microphones such as laptop or conference room microphones.
transcription: Optional[AudioTranscription]Configuration of the transcription model.
Configuration of the transcription model.
The language of the input audio. Supplying the input language in
ISO-639-1 (e.g. en) format
will improve accuracy and latency.
model: Optional[Union[str, Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more], null]]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
Literal["whisper-1", "gpt-4o-mini-transcribe", "gpt-4o-mini-transcribe-2025-12-15", 2 more]The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
The model to use for transcription. Current options are whisper-1, gpt-4o-mini-transcribe, gpt-4o-mini-transcribe-2025-12-15, gpt-4o-transcribe, and gpt-4o-transcribe-diarize. Use gpt-4o-transcribe-diarize when you need diarization with speaker labels.
An optional text to guide the model's style or continue a previous audio
segment.
For whisper-1, the prompt is a list of keywords.
For gpt-4o-transcribe models (excluding gpt-4o-transcribe-diarize), the prompt is a free text string, for example "expect words related to technology".
turn_detection: Optional[RealtimeTranscriptionSessionTurnDetection]Configuration for turn detection. Can be set to null to turn off. Server
VAD means that the model will detect the start and end of speech based on
audio volume and respond at the end of user speech.
Configuration for turn detection. Can be set to null to turn off. Server
VAD means that the model will detect the start and end of speech based on
audio volume and respond at the end of user speech.
Amount of audio to include before the VAD detected speech (in milliseconds). Defaults to 300ms.
Duration of silence to detect speech stop (in milliseconds). Defaults to 500ms. With shorter values the model will respond more quickly, but may jump in on short pauses from the user.
Activation threshold for VAD (0.0 to 1.0), this defaults to 0.5. A higher threshold will require louder audio to activate the model, and thus might perform better in noisy environments.
Type of turn detection, only server_vad is currently supported.
Expiration timestamp for the session, in seconds since epoch.
Additional fields to include in server outputs.
item.input_audio_transcription.logprobs: Include logprobs for input audio transcription.
The generated client secret value.
Create client secret
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ.get("OPENAI_API_KEY"), # This is the default and can be omitted
)
client_secret = client.realtime.client_secrets.create()
print(client_secret.expires_at){
"expires_at": 0,
"session": {
"client_secret": {
"expires_at": 0,
"value": "value"
},
"type": "realtime",
"audio": {
"input": {
"format": {
"rate": 24000,
"type": "audio/pcm"
},
"noise_reduction": {
"type": "near_field"
},
"transcription": {
"language": "language",
"model": "string",
"prompt": "prompt"
},
"turn_detection": {
"type": "server_vad",
"create_response": true,
"idle_timeout_ms": 5000,
"interrupt_response": true,
"prefix_padding_ms": 0,
"silence_duration_ms": 0,
"threshold": 0
}
},
"output": {
"format": {
"rate": 24000,
"type": "audio/pcm"
},
"speed": 0.25,
"voice": "ash"
}
},
"include": [
"item.input_audio_transcription.logprobs"
],
"instructions": "instructions",
"max_output_tokens": 0,
"model": "string",
"output_modalities": [
"text"
],
"prompt": {
"id": "id",
"variables": {
"foo": "string"
},
"version": "version"
},
"tool_choice": "none",
"tools": [
{
"description": "description",
"name": "name",
"parameters": {},
"type": "function"
}
],
"tracing": "auto",
"truncation": "auto"
},
"value": "value"
}Returns Examples
{
"expires_at": 0,
"session": {
"client_secret": {
"expires_at": 0,
"value": "value"
},
"type": "realtime",
"audio": {
"input": {
"format": {
"rate": 24000,
"type": "audio/pcm"
},
"noise_reduction": {
"type": "near_field"
},
"transcription": {
"language": "language",
"model": "string",
"prompt": "prompt"
},
"turn_detection": {
"type": "server_vad",
"create_response": true,
"idle_timeout_ms": 5000,
"interrupt_response": true,
"prefix_padding_ms": 0,
"silence_duration_ms": 0,
"threshold": 0
}
},
"output": {
"format": {
"rate": 24000,
"type": "audio/pcm"
},
"speed": 0.25,
"voice": "ash"
}
},
"include": [
"item.input_audio_transcription.logprobs"
],
"instructions": "instructions",
"max_output_tokens": 0,
"model": "string",
"output_modalities": [
"text"
],
"prompt": {
"id": "id",
"variables": {
"foo": "string"
},
"version": "version"
},
"tool_choice": "none",
"tools": [
{
"description": "description",
"name": "name",
"parameters": {},
"type": "function"
}
],
"tracing": "auto",
"truncation": "auto"
},
"value": "value"
}