xAI Grok Voice Transcribe 2.0: Speech-to-Text Model Release Notes and What's New

xAI released Grok Voice Transcribe 2.0 on September 18, 2026, a speech-to-text model the company describes as twice as accurate as Grok Voice Transcribe 1.0 at the same price. The release matters because it targets the hardest real-world audio — noisy telephony, multiple speakers, accents, and multilingual speech — while keeping existing Speech-to-Text API integrations working with no code changes.

What Grok Voice Transcribe 2.0 is

Grok Voice Transcribe 2.0 is xAI's latest speech-to-text model, built on the audio foundation model behind Grok Voice. According to xAI, Grok Voice already powers tens of thousands of customer-support calls per day, transcribes millions of hours of video narration, and runs voice agents in physical products, including the Grok assistant in Tesla vehicles.

The model is trained on what xAI describes as a unique dataset of live, noisy, multilingual audio recorded across a diverse set of environments, then refined with post-training. xAI positions the result as one of the most accurate transcription models for speech in real-world settings.

Accuracy: first among 32 streaming models on Artificial Analysis

On the public Artificial Analysis leaderboard, Grok Voice Transcribe 2.0 ranks first for accuracy among 32 streaming models. That is a public benchmark placement reported by xAI.

xAI also measures word error rate on four internal sets drawn from production traffic:

  • Telephony audio from customer-support calls
  • Conversations with Grok
  • Spoken credentials such as account codes and email addresses
  • Short multilingual voice commands

Grok Voice Transcribe 2.0 improves on Grok Voice Transcribe 1.0 across all four internal sets, and on telephony it leads every model xAI tested. xAI's framing is that most transcription models perform well on clean, single-speaker audio, while real-world audio is harder because of flaky phone lines, competing voices, local accents, and phone numbers or email addresses read aloud.

Multilingual transcription is the largest improvement

Grok Voice Transcribe 2.0 transcribes dozens of languages, detects the language automatically, and follows mid-recording language switches in a single pass. xAI states that multilingual accuracy is the model's largest improvement over Grok Voice Transcribe 1.0.

The hardest case is short phrases, such as in-car commands, which give the model little context for identifying the language. On xAI's short-phrase set, word error rate drops from 20.6% to 6.8% between version 1.0 and version 2.0.

Features and API capabilities

Grok Voice Transcribe 2.0 supports advanced configuration and controls. xAI states that existing Speech-to-Text API integrations receive the accuracy improvement with no code changes. The documented capabilities are:

  • Batch and streaming. Transcribe recorded files and URLs, or transcribe an audio stream in real time.
  • Word-level timestamps. Each word carries precise start and end times plus confidence scores.
  • Speaker diarization. Label each speaker in the transcript, at no additional cost.
  • Multichannel transcription. Transcribe up to 8 channels independently.
  • Key term biasing. Pass up to 100 domain terms per request, such as product names or medical vocabulary.
  • Text formatting. Numbers, dates, currencies, phone numbers, and email addresses are returned in written form.
  • Filler word removal. Omit fillers such as "um" and "uh" from the transcript.
  • Smart turn detection. Detect the end of a speaker's turn for voice agents.

Atlassian Loom uses Grok Voice Transcribe 2.0

Atlassian Loom, a widely used tool for recording and sharing screen recordings, found Grok Voice Transcribe 2.0 more accurate than its existing transcription solution. xAI describes the resulting workflow as accurate transcripts enabling new AI workflows: record an action plan in Loom, pipe the transcript into Cursor, and Cursor makes the code updates directly.

"We've always believed the best way to move work forward is to capture context once and let it flow everywhere. With Grok powering Loom's speech-to-text and Cursor turning that into code, we're closing the loop from context to code: record what you mean, and the work gets done. It's a glimpse of where AI-assisted development is headed."

— Sanchan Saxena, SVP of Teamwork Collection, Atlassian

Pricing and migration from version 1.0

Grok Voice Transcribe 2.0 pricing is identical to Grok Voice Transcribe 1.0:

  • Batch transcription: $0.10 per hour of audio
  • Streaming transcription: $0.20 per hour of audio
  • Diarization, timestamps, and key terms are included at no extra cost

Grok Voice Transcribe 2.0 will soon become the default in the Speech-to-Text API, and Grok Voice Transcribe 1.0 will be deprecated in the coming weeks. To remain on the older model during the transition, pin grok-voice-transcribe-1.0.

What the release signals

Grok Voice Transcribe 2.0 is a same-price accuracy upgrade rather than a new pricing tier, and xAI is pushing it to the default position in its Speech-to-Text API while deprecating version 1.0 within weeks. The emphasis on telephony, spoken credentials, and short multilingual commands indicates the model is aimed at production voice workloads — customer support, in-car assistants, and voice agents — rather than clean studio transcription. The Atlassian Loom integration illustrates the intended downstream pattern: transcription accuracy is treated as an input to agentic coding and workflow automation, not as an end product.

Sources