Realtime Beta session tokens

REST API endpoint to generate ephemeral session tokens for use in client-side applications.

Create session

post https://api.openai.com/v1/realtime/sessions

Create an ephemeral API token for use in client-side applications with the Realtime API. Can be configured with the same session parameters as the session.update client event.

It responds with a session object, plus a client_secret key which contains a usable ephemeral API token that can be used to authenticate browser clients for the Realtime API.

Request body

Ephemeral key returned by the API.

The format of input audio. Options are pcm16, g711_ulaw, or g711_alaw.

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 and should be treated as rough guidance rather than the representation understood by the model.

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.

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.

The set of modalities the model can respond with. To disable audio, set this to ["text"].

The format of output audio. Options are pcm16, g711_ulaw, or g711_alaw.

Reference to a prompt template and its variables. Learn more.

The speed of the model's spoken response. 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.

Sampling temperature for the model, limited to [0.6, 1.2]. Defaults to 0.8.

How the model chooses tools. Options are auto, none, required, or specify a function.

Tools (functions) available to the model.

Configuration options for tracing. 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.

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.

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.

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, and verse.

Returns

The created Realtime session object, plus an ephemeral key

Example request
1
2
3
4
5
6
7
8
curl -X POST https://api.openai.com/v1/realtime/sessions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-realtime",
    "modalities": ["audio", "text"],
    "instructions": "You are a friendly assistant."
  }'
Response
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
{
  "id": "sess_001",
  "object": "realtime.session",
  "model": "gpt-realtime-2025-08-25",
  "modalities": ["audio", "text"],
  "instructions": "You are a friendly assistant.",
  "voice": "alloy",
  "input_audio_format": "pcm16",
  "output_audio_format": "pcm16",
  "input_audio_transcription": {
      "model": "whisper-1"
  },
  "turn_detection": null,
  "tools": [],
  "tool_choice": "none",
  "temperature": 0.7,
  "max_response_output_tokens": 200,
  "speed": 1.1,
  "tracing": "auto",
  "client_secret": {
    "value": "ek_abc123", 
    "expires_at": 1234567890
  }
}

Create transcription session

post https://api.openai.com/v1/realtime/transcription_sessions

Create an ephemeral API token for use in client-side applications with the Realtime API specifically for realtime transcriptions. Can be configured with the same session parameters as the transcription_session.update client event.

It responds with a session object, plus a client_secret key which contains a usable ephemeral API token that can be used to authenticate browser clients for the Realtime API.

Request body

The set of items to include in the transcription. Current available items are: item.input_audio_transcription.logprobs

The format of input audio. Options are pcm16, g711_ulaw, or g711_alaw. For pcm16, input audio must be 16-bit PCM at a 24kHz sample rate, single channel (mono), and little-endian byte order.

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 transcription. The client can optionally set the language and prompt for transcription, these offer additional guidance to the transcription service.

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.

Returns

The created Realtime transcription session object, plus an ephemeral key

Example request
1
2
3
4
curl -X POST https://api.openai.com/v1/realtime/transcription_sessions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{}'
Response
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
{
  "id": "sess_BBwZc7cFV3XizEyKGDCGL",
  "object": "realtime.transcription_session",
  "modalities": ["audio", "text"],
  "turn_detection": {
    "type": "server_vad",
    "threshold": 0.5,
    "prefix_padding_ms": 300,
    "silence_duration_ms": 200
  },
  "input_audio_format": "pcm16",
  "input_audio_transcription": {
    "model": "gpt-4o-transcribe",
    "language": null,
    "prompt": ""
  },
  "client_secret": null
}

The session object

A Realtime session configuration object.

Configuration for input and output audio for the session.

Expiration timestamp for the session, in seconds since epoch.

Unique identifier for the session that looks like sess_1234567890abcdef.

Additional fields to include in server outputs.

null.

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.

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.

The Realtime model used for this session.

The object type. Always realtime.session.

The set of modalities the model can respond with. To disable audio, set this to ["text"].

How the model chooses tools. Options are auto, none, required, or specify a function.

Tools (functions) available to the model.

Configuration options for tracing. 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.

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.

OBJECT The session object
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
{
  "id": "sess_001",
  "object": "realtime.session",
  "expires_at": 1742188264,
  "model": "gpt-realtime",
  "output_modalities": ["audio"],
  "instructions": "You are a friendly assistant.",
  "tools": [],
  "tool_choice": "none",
  "max_output_tokens": "inf",
  "tracing": "auto",
  "truncation": "auto",
  "prompt": null,
  "audio": {
    "input": {
      "format": {
        "type": "audio/pcm",
        "rate": 24000
      },
      "transcription": { "model": "whisper-1" },
      "noise_reduction": null,
      "turn_detection": null
    },
    "output": {
      "format": {
        "type": "audio/pcm",
        "rate": 24000
      },
      "voice": "alloy",
      "speed": 1.0
    }
  }
}

The transcription session object

A new Realtime transcription session configuration.

When a session is created on the server via REST API, the session object also contains an ephemeral key. Default TTL for keys is 10 minutes. This property is not present when a session is updated via the WebSocket API.

Ephemeral key returned by the API. Only present when the session is created on the server via REST API.

The format of input audio. Options are pcm16, g711_ulaw, or g711_alaw.

Configuration of the transcription model.

The set of modalities the model can respond with. To disable audio, set this to ["text"].

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.

OBJECT The transcription session object
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
{
  "id": "sess_BBwZc7cFV3XizEyKGDCGL",
  "object": "realtime.transcription_session",
  "expires_at": 1742188264,
  "modalities": ["audio", "text"],
  "turn_detection": {
    "type": "server_vad",
    "threshold": 0.5,
    "prefix_padding_ms": 300,
    "silence_duration_ms": 200
  },
  "input_audio_format": "pcm16",
  "input_audio_transcription": {
    "model": "gpt-4o-transcribe",
    "language": null,
    "prompt": ""
  },
  "client_secret": null
}