{
  "@context": "https://schema.org",
  "@id": "https://aggr.aymericbeaumet.com/items/x-ai/2026-07-29-introducing-grok-voice-2-0/#webpage",
  "@type": [
    "WebPage",
    "ArchiveComponent"
  ],
  "articleBody": "Today, we're announcing Grok Voice 2.0, our next-generation voice model with improved intelligence, transcription accuracy, and conversational capabilities.\n\n## [Intelligence](https://x.ai/news/grok-voice-think-fast-2#intelligence)\n\nGrok Voice 2.0 is our most intelligent voice model yet, building on its predecessor with meaningful gains in speech reasoning, conversational ability, and tool use reliability.\n\nSpeech-to-speech benchmark comparison between Grok Voice 2.0 and competing models, per Artificial Analysis.\n| Benchmark | Grok Voice 2.0 (this release) | Grok Voice 1.0 | GPT-Realtime-2.1 (High) | Gemini 3.1 Flash (High) |\n| --------- | ----------------------------- | -------------- | ----------------------- | ----------------------- |\n| 82.9%     | 75.7%                         | 79.1%          | 69.5%                   |                         |\n| 97.2%     | 97.1%                         | 96.0%          | 96.6%                   |                         |\n| 95.1%     | 77.8%                         | 95.7%          | 74.3%                   |                         |\n| 56.5%     | 52.1%                         | 45.7%          | 37.7%                   |                         |\n| 0.70s     | 1.25s                         | —              | 2.98s                   |                         |\n\nSource: [Artificial Analysis](https://artificialanalysis.ai/speech-to-speech).\n\n## [Transcription Accuracy](https://x.ai/news/grok-voice-think-fast-2#transcription-accuracy)\n\nGrok Voice 2.0 outperforms even dedicated, state of the art transcription models when it comes to accuracy. In our evaluation across thousands of short phrases in 24 different languages, we've demonstrated a 1.5–2.0× improvement relative to Deepgram Nova 3 and ElevenLabs Scribe v2, and a 1.4× improvement relative to Grok Voice 1.0.\n\nThe gap between Grok Voice 2.0 and dedicated speech-to-text models widens to ~10× in noisy settings. We've focused on making Grok Voice 2.0 perform exceptionally well in real-world settings, with substantial background noise and telephony compression.\n\n### Transcription accuracy\n\nWord Error Rate (%). Lower is better.\n\nGrok Voice 2.0\n\nGrok Voice 1.0\n\nDeepgram Nova 3\n\nElevenLabs Scribe v2\n\n## [Reasoning Efficiency](https://x.ai/news/grok-voice-think-fast-2#reasoning-efficiency)\n\nGrok Voice models have a unique characteristic: they reason through queries while speaking. Reasoning in parallel with speech makes the model substantially smarter than other speech-to-speech models with no impact on latency.\n\nGrok Voice 2.0 has been trained to be very efficient with reasoning tokens relative to its predecessor. In production settings, this means that tool calls are snappier, usually executing before the end of the agent's first sentence.\n\n### Reasoning efficiency\n\nRelative reasoning tokens per response (P50). Lower is better.\n\nGrok Voice 2.0\n\n0.4×\n\nGrok Voice 1.0\n\n1.0×\n\n## [Conversational Capability](https://x.ai/news/grok-voice-think-fast-2#conversational-capability)\n\nWe've trained Grok Voice 2.0 to be a better conversationalist. By using extensive reinforcement learning to push the model towards patterns we see in real human conversation, we've found that broadly the model speaks in shorter sentences, asks one question at a time, and avoids fluff. From the user's perspective, conversations with the agent are simple and fluid, even though the model is often guiding the conversation through complex workflows and thinking several steps ahead behind the scenes.\n\n## [Migration & Pricing](https://x.ai/news/grok-voice-think-fast-2#migration--pricing)\n\nGrok Voice 2.0 is expected to result in improved performance across almost all use cases without any edits to existing prompts. In A/B testing this model on Starlink (+1 888 GO STARLINK), we've seen a significant increase in sales conversion rate and support containment rate.\n\nOn August 5, 2026, `grok-voice-latest` will move from Grok Voice 1.0 to Grok Voice 2.0 (`grok-voice-2.0`). No action needed to upgrade.\n\nWe believe pricing should be predictable and transparent. Grok Voice 2.0 is priced at [$0.08 / min of audio](https://docs.x.ai/developers/pricing?campaign=think-fast-2-blog#voice-api-pricing).\n\n## Start building with Grok Voice\n\n[Try It Free](https://console.x.ai/team/default/voice/agents?campaign=think-fast-2-blog&utm_source=website&utm_medium=referral&utm_campaign=think-fast-2-blog) [View Docs](https://docs.x.ai/developers/model-capabilities/audio/speech-to-speech?campaign=think-fast-2-blog)\n",
  "authors": [],
  "captured_at": "2026-10-19T01:26:53.998049939+00:00",
  "category": "ai",
  "content_html": "<p>Today, we're announcing Grok Voice 2.0, our next-generation voice model with improved intelligence, transcription accuracy, and conversational capabilities.</p>\n<h2><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://x.ai/news/grok-voice-think-fast-2#intelligence\">Intelligence</a></h2>\n<p>Grok Voice 2.0 is our most intelligent voice model yet, building on its predecessor with meaningful gains in speech reasoning, conversational ability, and tool use reliability.</p>\n<p>Speech-to-speech benchmark comparison between Grok Voice 2.0 and competing models, per Artificial Analysis.</p>\n<table>\n<thead>\n<tr>\n<th>Benchmark</th>\n<th>Grok Voice 2.0 (this release)</th>\n<th>Grok Voice 1.0</th>\n<th>GPT-Realtime-2.1 (High)</th>\n<th>Gemini 3.1 Flash (High)</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>82.9%</td>\n<td>75.7%</td>\n<td>79.1%</td>\n<td>69.5%</td>\n<td></td>\n</tr>\n<tr>\n<td>97.2%</td>\n<td>97.1%</td>\n<td>96.0%</td>\n<td>96.6%</td>\n<td></td>\n</tr>\n<tr>\n<td>95.1%</td>\n<td>77.8%</td>\n<td>95.7%</td>\n<td>74.3%</td>\n<td></td>\n</tr>\n<tr>\n<td>56.5%</td>\n<td>52.1%</td>\n<td>45.7%</td>\n<td>37.7%</td>\n<td></td>\n</tr>\n<tr>\n<td>0.70s</td>\n<td>1.25s</td>\n<td>—</td>\n<td>2.98s</td>\n<td></td>\n</tr>\n</tbody>\n</table>\n<p>Source: <a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://artificialanalysis.ai/speech-to-speech\">Artificial Analysis</a>.</p>\n<h2><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://x.ai/news/grok-voice-think-fast-2#transcription-accuracy\">Transcription Accuracy</a></h2>\n<p>Grok Voice 2.0 outperforms even dedicated, state of the art transcription models when it comes to accuracy. In our evaluation across thousands of short phrases in 24 different languages, we've demonstrated a 1.5–2.0× improvement relative to Deepgram Nova 3 and ElevenLabs Scribe v2, and a 1.4× improvement relative to Grok Voice 1.0.</p>\n<p>The gap between Grok Voice 2.0 and dedicated speech-to-text models widens to ~10× in noisy settings. We've focused on making Grok Voice 2.0 perform exceptionally well in real-world settings, with substantial background noise and telephony compression.</p>\n<h3>Transcription accuracy</h3>\n<p>Word Error Rate (%). Lower is better.</p>\n<p>Grok Voice 2.0</p>\n<p>Grok Voice 1.0</p>\n<p>Deepgram Nova 3</p>\n<p>ElevenLabs Scribe v2</p>\n<h2><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://x.ai/news/grok-voice-think-fast-2#reasoning-efficiency\">Reasoning Efficiency</a></h2>\n<p>Grok Voice models have a unique characteristic: they reason through queries while speaking. Reasoning in parallel with speech makes the model substantially smarter than other speech-to-speech models with no impact on latency.</p>\n<p>Grok Voice 2.0 has been trained to be very efficient with reasoning tokens relative to its predecessor. In production settings, this means that tool calls are snappier, usually executing before the end of the agent's first sentence.</p>\n<h3>Reasoning efficiency</h3>\n<p>Relative reasoning tokens per response (P50). Lower is better.</p>\n<p>Grok Voice 2.0</p>\n<p>0.4×</p>\n<p>Grok Voice 1.0</p>\n<p>1.0×</p>\n<h2><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://x.ai/news/grok-voice-think-fast-2#conversational-capability\">Conversational Capability</a></h2>\n<p>We've trained Grok Voice 2.0 to be a better conversationalist. By using extensive reinforcement learning to push the model towards patterns we see in real human conversation, we've found that broadly the model speaks in shorter sentences, asks one question at a time, and avoids fluff. From the user's perspective, conversations with the agent are simple and fluid, even though the model is often guiding the conversation through complex workflows and thinking several steps ahead behind the scenes.</p>\n<h2><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://x.ai/news/grok-voice-think-fast-2#migration--pricing\">Migration &amp; Pricing</a></h2>\n<p>Grok Voice 2.0 is expected to result in improved performance across almost all use cases without any edits to existing prompts. In A/B testing this model on Starlink (+1 888 GO STARLINK), we've seen a significant increase in sales conversion rate and support containment rate.</p>\n<p>On August 5, 2026, <code>grok-voice-latest</code> will move from Grok Voice 1.0 to Grok Voice 2.0 (<code>grok-voice-2.0</code>). No action needed to upgrade.</p>\n<p>We believe pricing should be predictable and transparent. Grok Voice 2.0 is priced at <a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://docs.x.ai/developers/pricing?campaign=think-fast-2-blog#voice-api-pricing\">$0.08 / min of audio</a>.</p>\n<h2>Start building with Grok Voice</h2>\n<p><a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://console.x.ai/team/default/voice/agents?campaign=think-fast-2-blog&amp;utm_source=website&amp;utm_medium=referral&amp;utm_campaign=think-fast-2-blog\">Try It Free</a> <a target=\"_blank\" rel=\"noopener noreferrer\" href=\"https://docs.x.ai/developers/model-capabilities/audio/speech-to-speech?campaign=think-fast-2-blog\">View Docs</a></p>\n",
  "content_markdown": "Today, we're announcing Grok Voice 2.0, our next-generation voice model with improved intelligence, transcription accuracy, and conversational capabilities.\n\n## [Intelligence](https://x.ai/news/grok-voice-think-fast-2#intelligence)\n\nGrok Voice 2.0 is our most intelligent voice model yet, building on its predecessor with meaningful gains in speech reasoning, conversational ability, and tool use reliability.\n\nSpeech-to-speech benchmark comparison between Grok Voice 2.0 and competing models, per Artificial Analysis.\n| Benchmark | Grok Voice 2.0 (this release) | Grok Voice 1.0 | GPT-Realtime-2.1 (High) | Gemini 3.1 Flash (High) |\n| --------- | ----------------------------- | -------------- | ----------------------- | ----------------------- |\n| 82.9%     | 75.7%                         | 79.1%          | 69.5%                   |                         |\n| 97.2%     | 97.1%                         | 96.0%          | 96.6%                   |                         |\n| 95.1%     | 77.8%                         | 95.7%          | 74.3%                   |                         |\n| 56.5%     | 52.1%                         | 45.7%          | 37.7%                   |                         |\n| 0.70s     | 1.25s                         | —              | 2.98s                   |                         |\n\nSource: [Artificial Analysis](https://artificialanalysis.ai/speech-to-speech).\n\n## [Transcription Accuracy](https://x.ai/news/grok-voice-think-fast-2#transcription-accuracy)\n\nGrok Voice 2.0 outperforms even dedicated, state of the art transcription models when it comes to accuracy. In our evaluation across thousands of short phrases in 24 different languages, we've demonstrated a 1.5–2.0× improvement relative to Deepgram Nova 3 and ElevenLabs Scribe v2, and a 1.4× improvement relative to Grok Voice 1.0.\n\nThe gap between Grok Voice 2.0 and dedicated speech-to-text models widens to ~10× in noisy settings. We've focused on making Grok Voice 2.0 perform exceptionally well in real-world settings, with substantial background noise and telephony compression.\n\n### Transcription accuracy\n\nWord Error Rate (%). Lower is better.\n\nGrok Voice 2.0\n\nGrok Voice 1.0\n\nDeepgram Nova 3\n\nElevenLabs Scribe v2\n\n## [Reasoning Efficiency](https://x.ai/news/grok-voice-think-fast-2#reasoning-efficiency)\n\nGrok Voice models have a unique characteristic: they reason through queries while speaking. Reasoning in parallel with speech makes the model substantially smarter than other speech-to-speech models with no impact on latency.\n\nGrok Voice 2.0 has been trained to be very efficient with reasoning tokens relative to its predecessor. In production settings, this means that tool calls are snappier, usually executing before the end of the agent's first sentence.\n\n### Reasoning efficiency\n\nRelative reasoning tokens per response (P50). Lower is better.\n\nGrok Voice 2.0\n\n0.4×\n\nGrok Voice 1.0\n\n1.0×\n\n## [Conversational Capability](https://x.ai/news/grok-voice-think-fast-2#conversational-capability)\n\nWe've trained Grok Voice 2.0 to be a better conversationalist. By using extensive reinforcement learning to push the model towards patterns we see in real human conversation, we've found that broadly the model speaks in shorter sentences, asks one question at a time, and avoids fluff. From the user's perspective, conversations with the agent are simple and fluid, even though the model is often guiding the conversation through complex workflows and thinking several steps ahead behind the scenes.\n\n## [Migration & Pricing](https://x.ai/news/grok-voice-think-fast-2#migration--pricing)\n\nGrok Voice 2.0 is expected to result in improved performance across almost all use cases without any edits to existing prompts. In A/B testing this model on Starlink (+1 888 GO STARLINK), we've seen a significant increase in sales conversion rate and support containment rate.\n\nOn August 5, 2026, `grok-voice-latest` will move from Grok Voice 1.0 to Grok Voice 2.0 (`grok-voice-2.0`). No action needed to upgrade.\n\nWe believe pricing should be predictable and transparent. Grok Voice 2.0 is priced at [$0.08 / min of audio](https://docs.x.ai/developers/pricing?campaign=think-fast-2-blog#voice-api-pricing).\n\n## Start building with Grok Voice\n\n[Try It Free](https://console.x.ai/team/default/voice/agents?campaign=think-fast-2-blog&utm_source=website&utm_medium=referral&utm_campaign=think-fast-2-blog) [View Docs](https://docs.x.ai/developers/model-capabilities/audio/speech-to-speech?campaign=think-fast-2-blog)\n",
  "dateCreated": "2026-10-19T01:26:53.998049939+00:00",
  "date_created": "2026-07-29T00:00:00+00:00",
  "date_modified": null,
  "date_published": "2026-07-29T00:00:00+00:00",
  "description": "Introducing our most capable speech-to-speech voice model.",
  "external_url": "https://x.ai/news/grok-voice-think-fast-2",
  "git": {
    "commit": "ebb0941c5c06a2ab2377342d607535262a3e969e",
    "path": "items/x-ai/2026/07/2026-07-29-introducing-grok-voice-2-0.md",
    "permalink": "https://github.com/aymericbeaumet/aggr-instance/blob/ebb0941c5c06a2ab2377342d607535262a3e969e/items/x-ai/2026/07/2026-07-29-introducing-grok-voice-2-0.md"
  },
  "id": "urn:aggr:item:7b4ec255d2658cd9e69d25fd91d5d40cdb87b7b9",
  "isBasedOn": {
    "@id": "https://x.ai/news/grok-voice-think-fast-2"
  },
  "mainEntity": {
    "@id": "https://x.ai/news/grok-voice-think-fast-2",
    "@type": "CreativeWork",
    "archivedAt": {
      "@id": "https://aggr.aymericbeaumet.com/items/x-ai/2026-07-29-introducing-grok-voice-2-0/#webpage"
    },
    "datePublished": "2026-07-29T00:00:00+00:00",
    "headline": "Introducing Grok Voice 2.0",
    "image": {
      "@type": "ImageObject",
      "caption": null,
      "height": 90,
      "url": "https://media.x.ai/v1/website/grok-voice-think-fast-2-19ad1e0b.webp",
      "width": 160
    },
    "timeRequired": "PT3M",
    "url": "https://x.ai/news/grok-voice-think-fast-2",
    "wordCount": 487
  },
  "name": "Archived snapshot: Introducing Grok Voice 2.0",
  "preview": {
    "alt": null,
    "color": null,
    "height": 90,
    "placeholder": {
      "data_url": "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNk+A8AAQUBAScY42YAAAAASUVORK5CYII=",
      "hash": ""
    },
    "url": "https://media.x.ai/v1/website/grok-voice-think-fast-2-19ad1e0b.webp",
    "width": 160
  },
  "reading_minutes": 3,
  "replicated_at": null,
  "source": "x-ai",
  "source_name": "News: Research, Product & Company Updates | SpaceXSI",
  "summary": "Introducing our most capable speech-to-speech voice model.",
  "tags": [],
  "temporalCoverage": "2026-10-19T01:26:53.998049939+00:00",
  "title": "Introducing Grok Voice 2.0",
  "url": "https://aggr.aymericbeaumet.com/items/x-ai/2026-07-29-introducing-grok-voice-2-0/",
  "word_count": 487
}