Grok Voice Think Fast 2.0 now available on AI Gateway
Grok Voice Think Fast 2.0 ships on Vercel's AI Gateway—parallel reasoning cuts latency for voice agents.
Why it matters
A production-ready speech-to-speech model with reasoning-while-speaking capability is now accessible via standard API infrastructure. Practitioners building voice agents gain a lower-latency, more efficient alternative; the parallel-reasoning architecture matters for real-time agent interaction.
The key facts
13 to knowGrok Voice Think Fast 2.0 released on Vercel AI Gateway
Speech-to-speech model with parallel reasoning (thinks while talking, no added latency)
Improved transcription accuracy in real-world conditions (background noise, telephony compression)
Reduced reasoning token usage allows tool calls to fire sooner, often mid-first sentence
Accessed via xAI realtime API with secure token minting via AI SDK
Playground available for testing; documentation includes voice agent build guide
Grok Voice Think Fast 2.0 released on AI Gateway
Speech-to-speech model with parallel reasoning (thinks while talking, no latency penalty)
Improved transcription accuracy in background noise and telephony compression
Reduced reasoning token count enables faster tool calls (often before agent's first sentence ends)
Available via xAI/grok-voice-think-fast-2.0 on realtime API
Accessible through AI SDK's realtime API and Vercel AI Gateway playground
Use case: voice agents with sub-second tool invocation
Go to the source
Vercel Blogvercel.com
Publisher excerpt: is now available on AI Gateway. It is a speech-to-speech voice model that takes audio in and audio out, improving on the previous Grok Voice model in reasoning, transcription accuracy, and conversation.Grok Voice Think Fast 2.0 from xAI The model reasons in parallel with speech, so it can think…