Grok Voice 3.0: an AI receptionist for UK small businesses?
xAI's Grok Voice 3.0 is a speech-to-speech model built for voice agents. Here's what it offers and where it could fit a UK small business.

Short answer: On 14 October 2026 xAI announced Grok Voice 3.0, a speech-to-speech model for voice agents that answer calls and hold spoken conversations. xAI says it replies in under a second and it costs $0.08 a minute in the Voice Agent API, the same as Grok Voice 2.0. For UK small businesses, the realistic uses are phone support, bookings and simple sales calls, as long as a human stays on hand.
xAI announced Grok Voice 3.0 on 14 October 2026 and calls it "the world's most intelligent speech-to-speech model". That's a big claim, so below we stick to what xAI has actually published: what the model does, what it costs, where it's already in use, and what a small business should think about before putting a voice agent on its phone line.
What xAI announced
Grok Voice 3.0 is a speech-to-speech model. It listens to someone speaking and answers out loud, rather than turning speech into text, writing a reply and reading it back. xAI built it for voice agents: software that holds a spoken conversation, takes a call or carries out a task. In other words, it's an AI voice agent model, and it's the same technology behind Grok voice mode.
According to xAI's announcement:
It's available now in xAI's Voice Agent API.
It replies in under one second, which xAI says is the fastest in its class.
Existing users moved over automatically. Agents using
grok-voice-latestwere switched to 3.0 with no changes needed.The price hasn't changed: $0.08 a minute ($4.80 an hour), the same as Grok Voice 2.0.
What it offers: the published results
xAI points to independent testing by Artificial Analysis, dated 13 October 2026. There, Grok Voice 3.0 ranks first on the Speech-to-Speech Index with a score of 86.9, up from 81.3 for Grok Voice 2.0. It also scores 97% on Full Duplex Bench, which measures how naturally a model handles pauses, turn-taking and interruptions.
On τ-Voice, a test of support calls with varied accents, background noise and interruptions, it scores 71.6%. xAI says that's comparable with OpenAI's GPT-Live-1 (Astra), but with 40% lower latency. On ServiceNow's EVA-Bench for enterprise voice agents, xAI reports the highest accuracy score, at 74%.
These are benchmark results, not a guarantee of how it will sound on your phone line. They do tell you the model was tested on calls with mixed accents, background noise and interruptions, not just clean studio audio.
Where it's already used
xAI says Grok Voice handles millions of conversations a day in Tesla vehicles. Drivers use it hands-free for simple jobs like checking a calendar or answering an email, and for bigger tasks like ordering a coffee or booking dinner. It also powers voice mode in Grok Bot.
The most useful example for businesses is Starlink. xAI says Starlink's call centres use Grok Voice to answer 20,000 customer calls a day across multiple languages. Roughly 70% are resolved with no human involved, and one in four sales calls ends in a purchase before the caller hangs up. A single agent verifies customers, troubleshoots hardware, issues credits and ships replacement equipment.
Using Grok Voice 3 as an AI receptionist for AI call answering
For an AI receptionist for small business use, Grok Voice 3 can act as an AI phone agent that picks up calls, answers questions and takes bookings. If your phone line and Google Business Profile drive bookings, see our local SEO services.
Starlink has engineers and a large budget behind its set-up. A small business won't get the same results overnight, but the same building blocks apply at a smaller scale:
Customer service out of hours. Answer common questions (opening times, delivery, returns) when nobody is in.
Bookings and enquiries. Take appointment requests for a salon, clinic, garage or consultancy, then pass them to your team to confirm.
First-line sales calls. Qualify an enquiry and collect contact details before a person calls back.
Accessibility. Give customers who find forms or small screens hard a way to talk to you instead.
Content and prototyping. Test spoken scripts, FAQs or demo calls before you record anything.
Who it suits, and who should wait
It suits businesses with a steady flow of similar calls and a developer, or an agency, who can connect it to booking or order systems through the API. At $4.80 an hour of conversation, the maths is easy to check against your call volume.
It suits you less if your calls are rare, sensitive or regulated. Immigration, legal, health and financial advice need a qualified person, and an automated agent should only route those calls, never advise.
Limitations and what to watch
Pricing is in US dollars, so your sterling cost moves with the exchange rate.
Benchmarks are averages. Test it with your own callers, accents and background noise first.
xAI hasn't published a full language list in the announcement. It says only that Starlink uses it "across multiple languages".
Data protection still applies. If a voice agent records or handles personal data, check UK GDPR, tell callers they're speaking to an automated system, and agree where recordings are stored.
Keep a way out. Always give callers a route to a human.
What it means for SEO, AEO and marketing
Voice agents answer from what you give them. The same is true of AI search, which is where SEO and AEO come in: tools like Google's AI Overviews and chat assistants quote pages that answer questions clearly. If your website has clear, accurate answers on prices, opening hours, services and areas covered, you've already written most of what a voice agent needs, and you've made your site easier for answer engines to quote.
So the practical step is the same either way: keep your FAQs, Google Business Profile, service pages and social media management accurate and consistent. Wrong opening hours on your website become wrong answers on the phone.
Grok Voice 3.0 and AI receptionist FAQs
What is Grok Voice 3.0?
A speech-to-speech model from xAI, announced on 14 October 2026, built for voice agents that hold spoken conversations, take calls and complete tasks.
How much does Grok Voice 3.0 cost?
xAI lists it at $0.08 a minute ($4.80 an hour) in its Voice Agent API, the same price as Grok Voice 2.0.
Do I need to change anything if I already use Grok Voice?
No, according to xAI. Agents using grok-voice-latest were switched to 3.0 automatically.
Can a small business use an AI receptionist?
Yes. An AI receptionist built on a voice agent such as Grok Voice 3 can answer calls, take messages and book appointments around the clock. Most small businesses have a developer or agency set it up, test it on real calls first and keep a human available for complex queries.
Thinking about a voice agent, or want your website ready for AI search? Get in touch with Click Katha and we'll talk it through.
Related reading: Grok 4 Heavy explained and ChatGPT Agent Mode for businesses.