Independent guide. Not affiliated with SpaceXSI, xAI or X.
Grok Voice for developers

Grok Voice API: pricing, features and voices for developers

What the Grok voice API costs per minute, what it supports, how it has changed since launch and how voice cloning works.

Checked October 18, 2026By GrokMag editorsIndependent · not affiliated with SpaceXSI, xAI or X
Short answer

The Grok speech-to-speech API (grok-voice-3.0) costs $0.08 per minute. Speech-to-text is $0.10 per hour (batch) or $0.20 (streaming), and text-to-speech is $15 per million characters. It offers 28 built-in voices, function calling, web and X search, and custom voices.

Pricing (October 2026)

ServicePrice
Speech-to-speech (grok-voice-3.0)$0.08 per minute ($4.80 per hour), plus $0.004 per text input event
Speech-to-text, batch$0.10 per hour
Speech-to-text, streaming$0.20 per hour
Text-to-speech$15 per 1 million characters

From the official developer pricing page, checked October 18, 2026. The voice agent API launched at $0.05 per minute in December 2025; the per-minute price has since risen to $0.08.

Timeline

  • Dec 17, 2025: Grok Voice Agent API launches for real-time voice apps.
  • May 2026: custom voices (voice cloning) arrive with Grok 4.3, from about a minute of speech plus a spoken consent check.
  • Jul 29, 2026: Grok Voice Think Fast 2.0 announced.
  • Oct 14, 2026: Grok Voice 3.0 released; it topped Artificial Analysis’s speech-to-speech index at launch.

What the API supports

  • A real-time WebSocket endpoint with automatic turn detection or push-to-talk
  • Function calling, plus built-in web search, X search and remote MCP tools
  • 28 built-in voices; custom voices from a reference clip of up to 120 seconds
  • Speech-to-text in 38+ languages; text-to-speech with inline tags such as laughs, whispers and pauses, and phone-line audio output
  • Up to 10 concurrent sessions per team and sessions up to 120 minutes
Use cloning responsibly. Only clone voices you have clear permission to use. The provider says its consent check blocks cloning other people, but no independent test results have been published.

Questions people ask

How much does the Grok voice API cost?

Speech-to-speech is $0.08 per minute as of October 2026, up from $0.05 at launch in December 2025.

Can the Grok API clone voices?

Yes, since May 2026, from a short reference recording with a spoken consent check.

What is the latest Grok voice model?

Grok Voice 3.0, released October 14, 2026.