{"id":248,"date":"2026-10-16T10:06:43","date_gmt":"2026-10-16T10:06:43","guid":{"rendered":"https:\/\/scoy.ai\/guides\/grok-voice-badge\/"},"modified":"2026-10-20T08:01:52","modified_gmt":"2026-10-20T08:01:52","slug":"grok-voice-badge","status":"publish","type":"post","link":"https:\/\/scoy.ai\/guides\/grok-voice-badge\/","title":{"rendered":"When xAI Launched Grok Voice 3.0, It Left Up a No. 1 Badge Its Own Chart Doesn’t Back"},"content":{"rendered":"\n
\n
\n

xAI’s voice page, this morning:<\/strong> “#1 Tau Voice Leaderboard.” The same badge was there on October 2, when Grok Voice 2.0 was the model on sale.<\/p>\n<\/div>\n\n\n\n

\n

xAI’s own launch chart, October 14:<\/strong> Grok Voice 3.0 scores 71.6% on Tau Voice. OpenAI’s GPT-Live-1 scores 74.5%. Grok Voice 2.0 scores 56.5%.<\/p>\n<\/div>\n<\/div>\n\n\n\n

Don’t pick a phone agent off that badge. Test it on your own calls first, because the badge and the chart can’t both be right.<\/p>\n\n\n\n

The reviews did catch that Grok Voice 3.0 came second on Tau Voice, the test that drops a voice model into a fake support call with tools and a policy to follow. What they didn’t do is open xAI’s sales page, which still says first. The Wayback Machine’s October 2 copy<\/a> shows the identical badge from before the new model shipped. xAI’s own announcement<\/a> is more careful than its sales page. It calls the score “comparable with GPT-Live-1”.<\/p>\n\n\n\n

I don’t know if the badge points at some other Tau Voice ranking. If it does, the page doesn’t say which one.<\/p>\n\n\n\n

\n

Agents using grok-voice-latest have already been switched over automatically, with no changes needed.<\/p>\nxAI, Grok Voice 3.0 announcement, October 14, 2026<\/cite><\/blockquote>\n\n\n\n

That line reads like good news. For an owner it means the voice answering your phone changed on October 14 and nobody asked you. Is your agent set to that “latest” option? Ask whoever built it. Then call your own number and listen to how it handles a hard question.<\/p>\n\n\n\n

What the Phone Line Adds<\/h2>\n\n\n\n

The price didn’t move: 8 cents a minute, same as Grok Voice 2.0. It isn’t the whole bill, though. xAI’s own phone demo runs through Twilio, so the phone line runs a second meter. Here’s a made-up shop taking 20 calls a weekday at three minutes each. The volume is invented. The prices were read this morning, October 16, 2026, from xAI’s announcement and Twilio’s US voice price list<\/a>.<\/p>\n\n\n\n

Line<\/th>Math<\/th>Per month<\/th><\/tr><\/thead>
Grok Voice 3.0<\/td>1,320 minutes x $0.08<\/td>$105.60<\/td><\/tr>
Twilio, incoming calls on a local number<\/td>1,320 minutes x $0.0085<\/td>$11.22<\/td><\/tr>
Twilio Media Streams (carries the call audio to the AI)<\/td>1,320 minutes x $0.0044<\/td>$5.81<\/td><\/tr>
Twilio local number<\/td>flat<\/td>$1.15<\/td><\/tr>
Total, 440 calls<\/td><\/td>$123.78<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

So the phone side adds about 17% on top of the AI bill, and the whole thing lands near 28 cents a call. I don’t know how xAI counts a minute while a caller sits on hold, and that could move the total. How many calls does your shop miss on a busy afternoon? That’s the number to put in the first row.<\/p>\n\n\n\n

xAI also says Starlink’s call centers settle about 70% of 20,000 daily calls with no person involved. That’s one source, xAI itself, and nobody outside has checked it yet. Who’d listen to the calls your agent gets wrong?<\/p>\n\n\n\n

If a phone job is on your list, send me one, two, or five<\/a> things you’d hand off. One is plenty.<\/p>\n","protected":false},"excerpt":{"rendered":"

xAI’s voice page still claims No. 1 on Tau Voice, while its own Grok Voice 3.0 chart puts it second. What a small shop’s phone agent costs a month.<\/p>\n","protected":false},"author":1,"featured_media":253,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[],"class_list":["post-248","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-news"],"_links":{"self":[{"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/posts\/248","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/comments?post=248"}],"version-history":[{"count":1,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/posts\/248\/revisions"}],"predecessor-version":[{"id":254,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/posts\/248\/revisions\/254"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/media\/253"}],"wp:attachment":[{"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/media?parent=248"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/categories?post=248"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/scoy.ai\/guides\/wp-json\/wp\/v2\/tags?post=248"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}