Documentation 2026-07-22 21:00 UTC to 2026-07-23 16:58 UTC

2 2 

3# Custom Voices3# Custom Voices

4 4 

5Clone a voice from a short reference clip and use it anywhere a built-in voice works. Upload an audio sample and immediately start using it in our TTS and Voice Agent APIs.5Clone a voice from a short reference clip and use it anywhere a built-in voice works. Upload an audio sample and immediately start using it in our TTS and Speech to Speech APIs.

6 6 

7> [!WARNING]7> [!WARNING]

8>8>

436});436});

437```437```

438 438 

439The response is `{"deleted": true}`. After deletion, subsequent requests for the same `voice_id` return `404` and any TTS / Voice Agent calls referencing it will fail with an unknown-voice error.439The response is `{"deleted": true}`. After deletion, subsequent requests for the same `voice_id` return `404` and any TTS / Speech to Speech calls referencing it will fail with an unknown-voice error.

440 440 

441## Using a Custom Voice441## Using a Custom Voice

442 442 

490asyncio.run(stream_with_custom_voice("nlbqfwie"))490asyncio.run(stream_with_custom_voice("nlbqfwie"))

491```491```

492 492 

493### Voice Agent API493### Speech to Speech API

494 494 

495Set `voice` in the `session.update` message. See the [Voice Agent API docs](/developers/model-capabilities/audio/voice-agent) for the full session lifecycle.495Set `voice` in the `session.update` message. See the [Speech to Speech API docs](/developers/model-capabilities/audio/speech-to-speech) for the full session lifecycle.

496 496 

497```python customLanguage="pythonWithoutSDK"497```python customLanguage="pythonWithoutSDK"

498import asyncio498import asyncio

2 2 

3# Ephemeral Tokens3# Ephemeral Tokens

4 4 

5Ephemeral tokens provide secure, short-lived authentication for client-side applications. Use them when connecting to the [Voice Agent API](/developers/model-capabilities/audio/voice-agent) from browsers or mobile apps to avoid exposing your API key.5Ephemeral tokens provide secure, short-lived authentication for client-side applications. Use them when connecting to the [Speech to Speech API](/developers/model-capabilities/audio/speech-to-speech) from browsers or mobile apps to avoid exposing your API key.

6 6 

7## How It Works7## How It Works

8 8 

1#### Model Capabilities1#### Model Capabilities

2 2 

3# Voice Agent API3# Speech to Speech API

4 4 

5Build real-time voice applications powered by Grok. Stream audio and text bidirectionally via WebSocket for voice assistants, phone agents, and interactive voice systems.5Build real-time voice applications powered by Grok. Stream audio and text bidirectionally via WebSocket for voice assistants, phone agents, and interactive voice systems.

6 6 

7## Quick Start7## Quick Start

8 8 

9Connect to the Voice Agent API and start a conversation:9Connect to the Speech to Speech API and start a conversation:

10 10 

11```python customLanguage="pythonWithoutSDK"11```python customLanguage="pythonWithoutSDK"

12import asyncio12import asyncio

149| `instructions` | string | System prompt |149| `instructions` | string | System prompt |

150| `reasoning.effort` | `"high"` | `"none"` | optional | Controls whether the model uses reasoning. Defaults to `"high"`. Supported only with `grok-voice-latest` and `grok-voice-think-fast-1.0`. |150| `reasoning.effort` | `"high"` | `"none"` | optional | Controls whether the model uses reasoning. Defaults to `"high"`. Supported only with `grok-voice-latest` and `grok-voice-think-fast-1.0`. |

151| `voice` | string | Voice selection: any built-in voice (e.g. `eve`, the default) or a [custom voice ID](/developers/model-capabilities/audio/custom-voices) (see [Available Voices](#available-voices)) |151| `voice` | string | Voice selection: any built-in voice (e.g. `eve`, the default) or a [custom voice ID](/developers/model-capabilities/audio/custom-voices) (see [Available Voices](#available-voices)) |

152| `tools` | array | Tools available to the voice agent. Supports `file_search`, `web_search`, `x_search`, `mcp`, and `function` types. See [Using Tools](#using-tools-with-grok-voice-agent-api). |152| `tools` | array | Tools available to the voice agent. Supports `file_search`, `web_search`, `x_search`, `mcp`, and `function` types. See [Using Tools](#using-tools-with-grok-speech-to-speech-api). |

153| `turn_detection.type` | string | null | `"server_vad"` for automatic detection, `null` for manual text turns |153| `turn_detection.type` | string | null | `"server_vad"` for automatic detection, `null` for manual text turns |

154| `turn_detection.threshold` | number | optional | VAD activation threshold (0.1–0.9). Higher values require louder audio to trigger. Default: `0.85`. |154| `turn_detection.threshold` | number | optional | VAD activation threshold (0.1–0.9). Higher values require louder audio to trigger. Default: `0.85`. |

155| `turn_detection.silence_duration_ms` | number | optional | How long the user must be silent (in ms) before the server ends the turn (0–10000). Higher values let users pause longer without being cut off. |155| `turn_detection.silence_duration_ms` | number | optional | How long the user must be silent (in ms) before the server ends the turn (0–10000). Higher values let users pause longer without being cut off. |

169 169 

170## Available Voices170## Available Voices

171 171 

172The same roster of voices works across the Voice Agent API and Text to Speech API. Browse the full list with tone descriptions and samples in the [voice table](/developers/model-capabilities/audio/text-to-speech#voices), or fetch it programmatically via [`GET /v1/tts/voices`](/developers/rest-api-reference/inference/voice). Pass the lowercase voice ID as the `voice` parameter on `session.update`.172The same roster of voices works across the Speech to Speech API and Text to Speech API. Browse the full list with tone descriptions and samples in the [voice table](/developers/model-capabilities/audio/text-to-speech#voices), or fetch it programmatically via [`GET /v1/tts/voices`](/developers/rest-api-reference/inference/voice). Pass the lowercase voice ID as the `voice` parameter on `session.update`.

173 173 

174### Custom Voices174### Custom Voices

175 175 

535 535 

536## Supported Languages536## Supported Languages

537 537 

538The Voice Agent API supports 20+ languages with native-quality accents. The model automatically detects the input language and responds naturally in the same language — no configuration required.538The Speech to Speech API supports 20+ languages with native-quality accents. The model automatically detects the input language and responds naturally in the same language — no configuration required.

539 539 

540| Language | Code |540| Language | Code |

541|----------|------|541|----------|------|

632}));632}));

633```633```

634 634 

635## Using Tools with Grok Voice Agent API635## Using Tools with Grok Speech to Speech API

636 636 

637The Grok Voice Agent API supports various tools that can be configured in your session to enhance the capabilities of your voice agent. Tools can be configured in the `session.update` message.637The Grok Speech to Speech API supports various tools that can be configured in your session to enhance the capabilities of your voice agent. Tools can be configured in the `session.update` message.

638 638 

639### Available Tool Types639### Available Tool Types

640 640 

1373 1373 

1374## Best Practices1374## Best Practices

1375 1375 

1376This section outlines key recommendations for building low-latency, reliable, and natural-feeling voice experiences using the xAI Voice Agent API.1376This section outlines key recommendations for building low-latency, reliable, and natural-feeling voice experiences using the xAI Speech to Speech API.

1377 1377 

1378### Minimize perceived latency with parallel initialization1378### Minimize perceived latency with parallel initialization

1379 1379 

1494 1494 

1495## SIP phone calls1495## SIP phone calls

1496 1496 

1497Route PSTN, contact-center, or PBX calls into a Voice Agent API session. See [SIP Phone Calls](/developers/model-capabilities/audio/voice-agent/sip) for API integration with `CreatePhoneNumberV2`, call control, DTMF, and telephony provider examples.1497Route PSTN, contact-center, or PBX calls into a Speech to Speech API session. See [SIP Phone Calls](/developers/model-capabilities/audio/speech-to-speech/sip) for API integration with `CreatePhoneNumberV2`, call control, DTMF, and telephony provider examples.

1498 1498 

1499## Migrating from OpenAI Realtime1499## Migrating from OpenAI Realtime

1500 1500 

1501If you have an existing application built on the [OpenAI Realtime API](https://developers.openai.com/api/docs/guides/realtime-conversations), switching to the Grok Voice Agent API requires only a few changes: update the base URL, swap your API key, and choose a Grok voice model.1501If you have an existing application built on the [OpenAI Realtime API](https://developers.openai.com/api/docs/guides/realtime-conversations), switching to the Grok Speech to Speech API requires only a few changes: update the base URL, swap your API key, and choose a Grok voice model.

1502 1502 

1503### Step 1 — Update the Base URL and API Key1503### Step 1 — Update the Base URL and API Key

1504 1504 

1645 1645 

1646## OpenAI Realtime API Compatibility1646## OpenAI Realtime API Compatibility

1647 1647 

1648The Grok Voice Agent API is compatible with the [OpenAI Realtime API](https://developers.openai.com/api/docs/guides/realtime-conversations). Most OpenAI client libraries and SDKs work with the xAI endpoint by changing the base URL to `wss://api.x.ai/v1/realtime`. This section documents event naming differences and unsupported events.1648The Grok Speech to Speech API is compatible with the [OpenAI Realtime API](https://developers.openai.com/api/docs/guides/realtime-conversations). Most OpenAI client libraries and SDKs work with the xAI endpoint by changing the base URL to `wss://api.x.ai/v1/realtime`. This section documents event naming differences and unsupported events.

1649 1649 

1650### Event Naming Differences1650### Event Naming Differences

1651 1651 

1#### Voice Agent API1#### Speech to Speech API

2 2 

3# SIP Phone Calls3# SIP Phone Calls

4 4 

5SIP lets you route PSTN, contact-center, or PBX calls into a Voice Agent API session.5SIP lets you route PSTN, contact-center, or PBX calls into a Speech to Speech API session.

6 6 

7### 1. Register the phone number7### 1. Register the phone number

8 8 

45 45 

46Open `wss://api.x.ai/v1/realtime?call_id={call_id}` with your xAI API key. Then send `session.update` to configure the voice agent for this call, followed by `response.create` when the agent should begin speaking.46Open `wss://api.x.ai/v1/realtime?call_id={call_id}` with your xAI API key. Then send `session.update` to configure the voice agent for this call, followed by `response.create` when the agent should begin speaking.

47 47 

48After connecting, the WebSocket behaves like any other Voice Agent API session. The SIP caller's audio is bridged into the session, and assistant audio is played back to the caller.48After connecting, the WebSocket behaves like any other Speech to Speech API session. The SIP caller's audio is bridged into the session, and assistant audio is played back to the caller.

49 49 

50```python customLanguage="pythonWithoutSDK"50```python customLanguage="pythonWithoutSDK"

51import asyncio51import asyncio

121 121 

122## DTMF phone keypresses122## DTMF phone keypresses

123 123 

124When using the Voice Agent API over SIP, phone keypresses (DTMF tones) are automatically buffered and flushed to the model as text input. The client receives `input_audio_buffer.dtmf_event_received` events as an audit trail of each keypress.124When using the Speech to Speech API over SIP, phone keypresses (DTMF tones) are automatically buffered and flushed to the model as text input. The client receives `input_audio_buffer.dtmf_event_received` events as an audit trail of each keypress.

125 125 

126### Flush triggers126### Flush triggers

127 127 

110| `diarize` | boolean | `false` | | When `true`, enables speaker diarization. Each word in the response includes a `speaker` field (integer) identifying the detected speaker. |110| `diarize` | boolean | `false` | | When `true`, enables speaker diarization. Each word in the response includes a `speaker` field (integer) identifying the detected speaker. |

111| `keyterm` | string | | | A key term to bias transcription toward (e.g. product names, proper nouns). Repeat the field for multiple terms (e.g. `keyterm=Understand+The+Universe`). Max 100 terms, each up to 50 characters. |111| `keyterm` | string | | | A key term to bias transcription toward (e.g. product names, proper nouns). Repeat the field for multiple terms (e.g. `keyterm=Understand+The+Universe`). Max 100 terms, each up to 50 characters. |

112| `filler_words` | boolean | `false` | | When `true`, filler words (e.g. "uh", "um", "er") are included in the transcript. When `false` (default), filler words are automatically removed from the transcript text and the `words` array. |112| `filler_words` | boolean | `false` | | When `true`, filler words (e.g. "uh", "um", "er") are included in the transcript. When `false` (default), filler words are automatically removed from the transcript text and the `words` array. |

113| `vad_threshold` | number | `0.5` | | Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio segments scoring below the threshold are treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise. `0` disables the gate. |

113 114 

114† Either `file` or `url` must be provided.115† Either `file` or `url` must be provided.

115 116 

117Option fields should precede `file` in the multipart body — for streamable uploads, fields sent after `file` may be ignored.

118 

116### Example with text formatting119### Example with text formatting

117 120 

118```bash121```bash

210| `keyterm` | string | | A key term to bias transcription toward (e.g. product names, proper nouns). Repeat the parameter for multiple terms (e.g. `keyterm=Understand+The+Universe`). Max 100 terms, each up to 50 characters. |213| `keyterm` | string | | A key term to bias transcription toward (e.g. product names, proper nouns). Repeat the parameter for multiple terms (e.g. `keyterm=Understand+The+Universe`). Max 100 terms, each up to 50 characters. |

211| `smart_turn` | number | | End-of-turn detection threshold (0.0–1.0). When set, enables Smart Turn — an ML model predicts whether the speaker has finished their thought at each silence boundary. See [Smart Turn](#smart-turn). |214| `smart_turn` | number | | End-of-turn detection threshold (0.0–1.0). When set, enables Smart Turn — an ML model predicts whether the speaker has finished their thought at each silence boundary. See [Smart Turn](#smart-turn). |

212| `smart_turn_timeout` | integer | | Maximum silence duration (ms) before forcing `speech_final`, even when the Smart Turn model predicts the speaker hasn't finished. Range: 1–5000. Only applies when `smart_turn` is enabled. See [Smart Turn](#smart-turn). |215| `smart_turn_timeout` | integer | | Maximum silence duration (ms) before forcing `speech_final`, even when the Smart Turn model predicts the speaker hasn't finished. Range: 1–5000. Only applies when `smart_turn` is enabled. See [Smart Turn](#smart-turn). |

216| `vad_threshold` | number | `0.08` | Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio in chunks scoring below the threshold is treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise. `0` disables the gate. Does not affect endpointing or `speech_final` timing. |

213 217 

214### Server Events218### Server Events

215 219 

1298* [TTS Playground](https://console.x.ai/team/default/voice/text-to-speech?campaign=voice-docs-tts) - Try voices and speech tags in your browser1298* [TTS Playground](https://console.x.ai/team/default/voice/text-to-speech?campaign=voice-docs-tts) - Try voices and speech tags in your browser

1299* [Create an API Key](https://console.x.ai/team/default/api-keys?campaign=voice-docs-tts) - Get started with the API1299* [Create an API Key](https://console.x.ai/team/default/api-keys?campaign=voice-docs-tts) - Get started with the API

1300* [Voice Overview](/developers/model-capabilities/audio/voice) - Overview of all xAI voice capabilities1300* [Voice Overview](/developers/model-capabilities/audio/voice) - Overview of all xAI voice capabilities

1301* [Voice Agent API](/developers/model-capabilities/audio/voice-agent) - Real-time voice conversations via WebSocket1301* [Speech to Speech API](/developers/model-capabilities/audio/speech-to-speech) - Real-time voice conversations via WebSocket

1302* [API Reference](/developers/rest-api-reference/inference/voice#text-to-speech---rest) - Full TTS endpoint specification1302* [API Reference](/developers/rest-api-reference/inference/voice#text-to-speech---rest) - Full TTS endpoint specification

1303* [List Voices](/developers/rest-api-reference/inference/voice#text-to-speech---list-voices) - Programmatically discover available voices1303* [List Voices](/developers/rest-api-reference/inference/voice#text-to-speech---list-voices) - Programmatically discover available voices

4 4 

5The xAI Voice APIs offer a range of powerful voice capabilities, all powered by Grok, with enterprise-grade reliability and sub-second latency.5The xAI Voice APIs offer a range of powerful voice capabilities, all powered by Grok, with enterprise-grade reliability and sub-second latency.

6 6 

7## Voice Agent API7## Speech to Speech API

8 8 

9Build real-time, speech-to-speech voice agents over WebSockets, with low-latency turn-taking and tool use. For client-side apps, use [Ephemeral Tokens](/developers/model-capabilities/audio/ephemeral-tokens) to connect securely without exposing your API key.9Build real-time, speech-to-speech voice agents over WebSockets, with low-latency turn-taking and tool use. For client-side apps, use [Ephemeral Tokens](/developers/model-capabilities/audio/ephemeral-tokens) to connect securely without exposing your API key.

10 10 

260fs.writeFileSync("custom.mp3", Buffer.from(await speech.arrayBuffer()));260fs.writeFileSync("custom.mp3", Buffer.from(await speech.arrayBuffer()));

261```261```

262 262 

263The custom `voice_id` also works with the streaming TTS WebSocket and the Voice Agent realtime API. See the [Custom Voices guide](/developers/model-capabilities/audio/custom-voices) for the full API.263The custom `voice_id` also works with the streaming TTS WebSocket and the Speech to Speech realtime API. See the [Custom Voices guide](/developers/model-capabilities/audio/custom-voices) for the full API.

264 264 

265## Voices265## Voices

266 266 

267When using the Voice Agent API or Text to Speech, you can choose from the full set of built-in voices. Each has its own personality and tone, so pick the one that best fits your application (`eve` is the default):267When using the Speech to Speech API or Text to Speech, you can choose from the full set of built-in voices. Each has its own personality and tone, so pick the one that best fits your application (`eve` is the default):

268 268 

269### Enterprise Compliance & Security269### Enterprise Compliance & Security

270 270 

32 32 

33* `filler_words` ("true" | "false") — When \`true\`, filler words (e.g. "uh", "um", "er") are included in the transcript. When \`false\` (default), filler words are automatically removed from the transcript text and the \`words\` array.33* `filler_words` ("true" | "false") — When \`true\`, filler words (e.g. "uh", "um", "er") are included in the transcript. When \`false\` (default), filler words are automatically removed from the transcript text and the \`words\` array.

34 34 

35* `vad_threshold` (number) — Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio segments scoring below the threshold are treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise; \`0\` disables the gate entirely. Default: \`0.5\`.

36 

35### Response Body37### Response Body

36 38 

37* `text` (string, required) — Full transcript text. For multichannel requests, this is a merged transcript across all channels (words interleaved by timestamp).39* `text` (string, required) — Full transcript text. For multichannel requests, this is a merged transcript across all channels (words interleaved by timestamp).

168 170 

169* `smart_turn_timeout` (integer, optional) — Maximum silence duration in milliseconds before forcing \`speech\_final\`, even when the Smart Turn model predicts the speaker hasn't finished. Acts as a safety net to prevent sessions from hanging during extended silence. Only applies when \`smart\_turn\` is enabled. Range: 1–5000. Example: \`smart\_turn\_timeout=3000\`.171* `smart_turn_timeout` (integer, optional) — Maximum silence duration in milliseconds before forcing \`speech\_final\`, even when the Smart Turn model predicts the speaker hasn't finished. Acts as a safety net to prevent sessions from hanging during extended silence. Only applies when \`smart\_turn\` is enabled. Range: 1–5000. Example: \`smart\_turn\_timeout=3000\`.

170 172 

173* `vad_threshold` (number, optional, default: 0.08) — Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio in chunks scoring below the threshold is treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise; \`0\` disables the gate entirely. Does not affect endpointing or \`speech\_final\` timing. Default: \`0.08\`.

174 

171### Client Messages175### Client Messages

172 176 

173* `Binary frame (audio)` — Send raw audio as binary WebSocket frames in the encoding specified by the \`encoding\` query parameter. Audio should be streamed in real-time-paced chunks (e.g. 100 ms at a time). No base64 encoding — send raw bytes directly.177* `Binary frame (audio)` — Send raw audio as binary WebSocket frames in the encoding specified by the \`encoding\` query parameter. Audio should be streamed in real-time-paced chunks (e.g. 100 ms at a time). No base64 encoding — send raw bytes directly.

843 843 

844* `filler_words` ("true" | "false") — When \`true\`, filler words (e.g. "uh", "um", "er") are included in the transcript. When \`false\` (default), filler words are automatically removed from the transcript text and the \`words\` array.844* `filler_words` ("true" | "false") — When \`true\`, filler words (e.g. "uh", "um", "er") are included in the transcript. When \`false\` (default), filler words are automatically removed from the transcript text and the \`words\` array.

845 845 

846* `vad_threshold` (number) — Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio segments scoring below the threshold are treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise; \`0\` disables the gate entirely. Default: \`0.5\`.

847 

846### Response Body848### Response Body

847 849 

848* `text` (string, required) — Full transcript text. For multichannel requests, this is a merged transcript across all channels (words interleaved by timestamp).850* `text` (string, required) — Full transcript text. For multichannel requests, this is a merged transcript across all channels (words interleaved by timestamp).

979 981 

980* `smart_turn_timeout` (integer, optional) — Maximum silence duration in milliseconds before forcing \`speech\_final\`, even when the Smart Turn model predicts the speaker hasn't finished. Acts as a safety net to prevent sessions from hanging during extended silence. Only applies when \`smart\_turn\` is enabled. Range: 1–5000. Example: \`smart\_turn\_timeout=3000\`.982* `smart_turn_timeout` (integer, optional) — Maximum silence duration in milliseconds before forcing \`speech\_final\`, even when the Smart Turn model predicts the speaker hasn't finished. Acts as a safety net to prevent sessions from hanging during extended silence. Only applies when \`smart\_turn\` is enabled. Range: 1–5000. Example: \`smart\_turn\_timeout=3000\`.

981 983 

984* `vad_threshold` (number, optional, default: 0.08) — Speech-probability threshold for the voice-activity gate (0.0–1.0). Audio in chunks scoring below the threshold is treated as non-speech and skipped for transcription. Lower values transcribe quieter or noisier speech (e.g. narrowband telephony) but may produce spurious text for background noise; \`0\` disables the gate entirely. Does not affect endpointing or \`speech\_final\` timing. Default: \`0.08\`.

985 

982### Client Messages986### Client Messages

983 987 

984* `Binary frame (audio)` — Send raw audio as binary WebSocket frames in the encoding specified by the \`encoding\` query parameter. Audio should be streamed in real-time-paced chunks (e.g. 100 ms at a time). No base64 encoding — send raw bytes directly.988* `Binary frame (audio)` — Send raw audio as binary WebSocket frames in the encoding specified by the \`encoding\` query parameter. Audio should be streamed in real-time-paced chunks (e.g. 100 ms at a time). No base64 encoding — send raw bytes directly.

1047 1051 

1048### Response Body1052### Response Body

1049 1053 

1050* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Voice Agent \`session.update\` message.1054* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Speech to Speech \`session.update\` message.

1051 1055 

1052* `name` (string | null) — Display name.1056* `name` (string | null) — Display name.

1053 1057 

1156 1160 

1157* `voices` (array\<object>, required) — List of custom voices owned by the calling team.1161* `voices` (array\<object>, required) — List of custom voices owned by the calling team.

1158 1162 

1159 * `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Voice Agent \`session.update\` message.1163 * `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Speech to Speech \`session.update\` message.

1160 1164 

1161 * `name` (string | null) — Display name.1165 * `name` (string | null) — Display name.

1162 1166 

1253 1257 

1254### Response Body1258### Response Body

1255 1259 

1256* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Voice Agent \`session.update\` message.1260* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Speech to Speech \`session.update\` message.

1257 1261 

1258* `name` (string | null) — Display name.1262* `name` (string | null) — Display name.

1259 1263 

1349 1353 

1350### Response Body1354### Response Body

1351 1355 

1352* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Voice Agent \`session.update\` message.1356* `voice_id` (string, required) — 8-character lowercase alphanumeric voice identifier. Use this as \`voice\_id\` in \`POST /v1/tts\`, as the \`voice\` query parameter on the streaming TTS WebSocket, or as \`voice\` in a Speech to Speech \`session.update\` message.

1353 1357 

1354* `name` (string | null) — Display name.1358* `name` (string | null) — Display name.

1355 1359 

6 6 

7## SDK Support7## SDK Support

8 8 

9Remote MCP tools are supported in the xAI native SDK, the OpenAI compatible Responses API, and the [Voice Agent API](/developers/model-capabilities/audio/voice-agent#remote-mcp-tools).9Remote MCP tools are supported in the xAI native SDK, the OpenAI compatible Responses API, and the [Speech to Speech API](/developers/model-capabilities/audio/speech-to-speech#remote-mcp-tools).

10 10 

11> [!NOTE]11> [!NOTE]

12>12>