Grok Build adds /voice dictation with Ctrl+Space transcription
Grok Build added speech-to-text dictation for coding agents through /voice or Ctrl+Space. Try it to bring Grok-powered real-time voice input into CLI coding workflows.

TL;DR
- testingcatalog's demo shows Grok Build taking a spoken coding prompt through
/voice; Wes Roth's note saysCtrl+Spacealso opens real-time transcription. - Wes Roth's post frames the feature as speech-to-text for coding agents, not a general Grok chat voice mode.
- The same Grok Voice push now includes xAI's Voice Agent Builder, where testingcatalog's console screenshot shows beta templates for support, sales, scheduling, personal assistant, and lead qualification.
- ai_for_success's launch demo lists the broader voice-agent package: 25+ languages, 80+ built-in voices, connectors, MCP support, and pricing from $0.05/min.
You can watch the terminal demo, compare it with xAI's Voice Agent Builder launch, and browse the Grok Build command docs that have not yet added /voice to the public command table. xAI's pricing page lists standalone speech-to-text at $0.10/hour for REST and $0.20/hour for streaming, while realtime voice agents are priced at $0.05/min.
/voice in the terminal
Grok Build now accepts dictated coding instructions inside the CLI. The demo starts with /voice, opens a recording interface, transcribes the spoken request, and passes the resulting text into the coding agent.
According to Wes Roth's follow-up, the second activation path is Ctrl+Space, and the transcription is powered by Grok speech-to-text.
Ctrl+Space transcription
The activation surface is small enough to matter: one slash command and one keyboard shortcut.
/voice, typed into Grok Build.Ctrl+Space, used as a direct hotkey.- Real-time transcription, before the coding agent processes the instruction.
On xAI's Grok Build overview, the product is positioned as an extensible coding agent usable through an interactive TUI, headless scripts, bots, or Agent Client Protocol clients. Voice dictation adds another input path to that same terminal workflow.
Voice Agent Builder
The CLI dictation landed next to a larger Grok Voice rollout. xAI's official Voice Agent Builder post describes a beta no-code platform for production voice agents with telephony, knowledge retrieval, tools, guardrails, MCPs, observability, SIP number support, and WebSocket clients.
The console screenshot shows five starter templates:
- Customer Support
- Sales Associate
- Appointment Scheduler
- Personal Assistant
- Lead Qualification
xAI says Voice Agent Builder uses a speech-to-speech path built for Grok Voice, rather than a stack assembled from separate speech recognition, language model, and text-to-speech services.
Pricing and docs
The current xAI pricing docs list four voice meters:
- Realtime: $0.05/min, or $3.00/hour.
- Realtime text input: $0.004/message.
- Text to speech: $15.00 per 1M characters.
- Speech to text: $0.10/hour for REST, $0.20/hour for streaming.
The public Grok Build docs lag the social demos. The modes and commands page lists core TUI commands like /context, /model, /compact, /usage, /plugins, /skills, and /mcps, but not /voice; the Build changelog read by Exa topped out at v0.2.73 on June 28, before the voice-dictation posts surfaced.