Introducing Grok Voice Transcribe 2.0
If your organisation transcribes meetings, calls or training sessions, this is a meaningfully more accurate transcription model at no extra cost — useful for generating searchable meeting notes or captioning workshop recordings. It's an API-level change though, so you'd need a developer or a tool built on xAI's API to actually use it, not something you'd see directly in a consumer Grok app.
What changed
Source summary
xAI released Grok Voice Transcribe 2.0, a speech-to-text model it says is twice as accurate as version 1.0 at the same price ($0.10/hour batch, $0.20/hour streaming), with word-level timestamps and speaker diarisation for phone calls, meetings, videos and podcasts. Version 2.0 will become the API default in the coming weeks, with version 1.0 deprecated; users can pin grok-voice-transcribe-1.0 to keep the old behaviour during migration.
Why it matters to your work
Practical interpretation
If your organisation transcribes meetings, calls or training sessions, this is a meaningfully more accurate transcription model at no extra cost — useful for generating searchable meeting notes or captioning workshop recordings. It's an API-level change though, so you'd need a developer or a tool built on xAI's API to actually use it, not something you'd see directly in a consumer Grok app.
When this could help
Worth evaluating if you run transcription at scale and want to benchmark against your current provider, especially for telephony, multilingual, or noisy-environment audio where the accuracy gains are most pronounced.
Grok Voice Transcribe 2.0 claims twice the accuracy of 1.0 at the same price ($0.10/hr batch, $0.20/hr streaming), ranks first on the Artificial Analysis streaming accuracy leaderboard across 32 models, and includes speaker diarisation, word-level timestamps, 8-channel transcription, key term biasing, and automatic language detection at no extra cost.
How to get started
Existing API integrations get the accuracy gains without code changes once 2.0 becomes the default. To try it now, select grok-voice-transcribe-2.0 in the API or use the console.
Provider guidance ↗
Build it into your routine
SCALA suggested workflow
Transcription provider comparison
Run the same test audio (telephony, multilingual, noisy) through your current STT provider and Grok Voice Transcribe 2.0. Compare word error rate and cost per hour to decide if switching saves money or improves accuracy.
What to check
Sources & provider guides
Official documentation checked 12 Oct 2026. SCALA prompts and workflows are suggestions, not claims of hands-on testing.
Source & availability
This is a summary of a published provider announcement, with practical context from SCALA’s feed.
Plan, region, workspace and rollout conditions may affect access in Australia. Check the original source for current availability.
x.ai