xAI has announced Grok Voice Think Fast 2.0, a new speech-to-speech voice model the company says is its most capable yet, with gains in intelligence, transcription accuracy and conversational ability.

According to benchmarks xAI cites from Artificial Analysis, the new model scores 82.9% on the AA Speech-to-Speech Quality Index, ahead of its predecessor as well as rival models GPT-Realtime-2.1 and Gemini 3.1 Flash. xAI also reports transcription improvements of 1.5 to 2 times over Deepgram Nova 3 and ElevenLabs Scribe v2 across 24 languages, widening to around tenfold in noisy environments.

The model reasons while speaking rather than pausing to think, which xAI says makes it markedly smarter without adding latency, while using fewer reasoning tokens than its predecessor. The company points to A/B testing on Starlink's support line as evidence of improved sales conversion and support containment.

From 5 August 2026, grok-voice-latest will automatically move to Think Fast 2.0, priced at $0.08 per minute of audio. Users wanting to remain on the previous version must pin grok-voice-think-fast-1.0 beforehand.

AI Governance 2026: The Compliance Gamble Facing Every High-Risk AI Deployer
Are you building your EU AI Act compliance strategy on a foundation that was never designed for the purpose? With harmonized standards still in draft and the high-risk provisions approaching, organisations face a strategic gamble: bet on the standard arriving in time, or accept a double compliance burden. 80-90% of requirements are common across all pathways — but the remaining 10-20% carries material consequences. You’ll learn: • Why the EU AI Act QMS is fundamentally different from ISO 9001, 27001, or 42001 • What the four compliance pathways actually require — and their different burdens • Why ISO 42001 does not lead to EU AI Act compliance • How AI governance is being confused with AI management — and why it matters • A practical 90-day readiness plan from frontline practitioners • Why directors face personal, uninsurable liability for AI decisions Key topics: EU AI Act high-risk compliance • Quality Management Systems • PREN18286 • Harmonized standards • AI sovereignty • Shadow AI • Automation bias • Data integrity • AI literacy training • Director liability • Conformity assessment • Role-based competency Essential viewing for CISOs, CIOs, CFOs, Chief Legal Officers, and board members responsible for AI strategy, regulatory compliance, and risk management.
BrightTALK

Share this post
The link has been copied!