xAI Formally Opens Grok Voice API for STT and TTS Services

Speech-to-text (STT) and text-to-speech (TTS) are foundational technologies for voice assistants, customer service and accessibility applications. Elon Musk's xAI has spun off technology used by Grok Voice, Tesla vehicles and Starlink customer service into standalone APIs for developers. The move puts it in competition with ElevenLabs and Deepgram for enterprise voice business. Its TTS service also supports tags for controlling laughter, whispers and pauses.

xAI opened the APIs on April 17, 2026, pricing batch STT at $0.10 per hour and streaming STT at $0.20 per hour, while TTS cost $4.20 per million characters at the time. The company said its STT service had an overall error rate of 6.9%, outperforming ElevenLabs' 9.0%. On April 23, xAI also released Starlink performance data showing that Grok Voice autonomously resolved 70% of customer-service inquiries and converted 20% of sales calls.