Grok Voice Think Fast 1.0
Grok Voice Think Fast 1.0 is xAI's realtime voice model for speech-to-speech agents, streaming audio bidirectionally over WebSocket through Vercel AI Gateway.
import { gateway } from '@ai-sdk/gateway';
export async function POST() { const { token, url } = await gateway.experimental_realtime.getToken({ model: 'xai/grok-voice-think-fast-1.0', });
return Response.json({ token, url, tools: [] });}Playground
Try out Grok Voice Think Fast 1.0 by xAI. Usage is billed to your team at API rates. Free users (those who haven't made a payment) get $5 of credits every 30 days.
Grok Voice Think Fast 1.0Voice agent
Talk to a realtime agent. It listens to your voice and replies with audio.
Start the session and ask the agent something.
Providers
Route requests across multiple providers. Copy a provider slug to set your preference. Visit the docs for more info. Using a provider means you agree to their terms, listed under Legal.
More models by xAI
| Model |
|---|
| 500K | 1.8s | 56tps | $2/M | $6/M | Read: $0.3/M Write: — | $5/K + input costs | +3 | | 07/08/2026 | |||
| 256K | 0.9s | 130tps | $1/M | $2/M | Read: $0.2/M Write: — | $5/K + input costs | +2 | | 05/20/2026 | |||
| 1M | 0.7s | 146tps | $1.25/M | $2.50/M | Read: $0.2/M Write: — | $5/K + input costs | +3 |  | 04/30/2026 | |||
| 2M | 0.3s | 80tps | $1.25/M | $2.50/M | Read: $0.2/M Write: — | $5/K + input costs | +2 | | 03/10/2026 | |||
| 1M | 0.5s | 190tps | $0.20/M | $0.50/M | Read:$0.05/M Write:— | — | +1 | | 11/19/2025 | |||
| 1M | 2.8s | 130tps | $0.20/M | $0.50/M | Read:$0.05/M Write:— | — | +2 | | 11/19/2025 |
About Grok Voice Think Fast 1.0
Grok Voice Think Fast 1.0 is xAI's realtime voice model, released April 23, 2026 and available through AI Gateway. Grok Voice Think Fast 1.0 handles full-duplex, speech-to-speech conversation: audio streams in, and spoken replies stream back with low latency. Background reasoning lets Grok Voice Think Fast 1.0 think through hard queries while the conversation keeps moving. xAI positions the model for customer support, sales, and other high-stakes voice workflows.
Grok Voice Think Fast 1.0 captures structured data from natural speech, including email addresses, street addresses, phone numbers, and account numbers, even through accents and mid-sentence corrections. Tool calling works inside the session, so a voice agent can act on external systems mid-conversation. xAI reports support for conversations in more than 25 languages and robustness under telephony conditions like noise and interruptions.
Through AI Gateway, a server route mints a short-lived token with gateway.experimental_realtime.getToken, so your API key never reaches the client. In the browser, the useRealtime hook from @ai-sdk/react manages the WebSocket connection, microphone capture, and audio playback. In Node.js, the realtime model acts as a codec for a WebSocket you drive yourself. Session config covers voice, turn detection, instructions, and tools. Realtime support ships on the canary releases of the AI SDK, and you can try Grok Voice Think Fast 1.0 in the AI Gateway playground at https://ai-sdk.dev/playground/xai:grok-voice-think-fast-1.0.
What To Consider When Choosing a Provider
- Configuration: Grok Voice Think Fast 1.0 supports speech-to-speech only. Transcription and translation aren't handled in realtime sessions, so use
grok-sttfor recorded audio. Image input isn't supported either. - Configuration: AI Gateway enforces session limits: 25 minutes maximum duration, a five-minute idle timeout, and a cap on concurrent sessions per team. Reconnecting doesn't resume a previous session, so replay any context you need after a drop. Realtime support is experimental and ships on canary AI SDK releases, so expect API changes.
- Zero Data Retention: AI Gateway does not currently support Zero Data Retention for this model. See the documentation for models that support ZDR.
- Authentication: AI Gateway authenticates requests using an API key or OIDC token. You do not need to manage provider credentials directly.
When to Use Grok Voice Think Fast 1.0
Best for
- Voice Assistants and Agents: Full-duplex conversation with spoken replies in real time
- Phone and Support Lines: Robust handling of noise, accents, and interruptions on telephony audio
- Structured Data Capture: Collects emails, addresses, and account numbers from natural speech
- Tool-Calling Voice Workflows: Session tools let agents act on external systems mid-conversation
- Multilingual Voice Apps: Conversations in more than 25 languages
Consider alternatives when
- Recorded Audio Transcription:
grok-stttranscribes files, since realtime sessions are speech-to-speech only - One-Way Speech Output:
grok-ttsgenerates narration without a live session - Long Uninterrupted Sessions: AI Gateway caps realtime sessions at 25 minutes, so plan reconnect logic
Conclusion
Grok Voice Think Fast 1.0 brings xAI's realtime voice stack to AI Gateway. Speech streams both directions over WebSocket, with background reasoning, tool calling, and structured data capture built for production voice agents. Mint a token on the server, connect with the useRealtime hook, and see live pricing on this page.