Grok Voice Realtime: xAI’s Audio-to-Audio Model Explained

xAI introduced grok-voice/realtime, an audio-to-audio agent available via the fal platform (endpoint fal-ai/grok-voice). The model accepts speech recordings and returns a voiced response, transcript, and duration; bidirectional WebSocket streaming is claimed.

• Input: audio up to 10 minutes and 50 MB, converted to 16-bit PCM 24 kHz mono
• Prompt optional, up to 20,000 characters; default persona is Grok
• Tools: web search, X search, and MCP servers (off by default)
• Price, latency, and availability not disclosed