whatmodel.sh / stt / Grok Voice Transcribe 1.0
Grok Voice Transcribe 1.0
↗ share
+ contribute
Grok Voice Transcribe 1.0 is a closed-weights speech-to-text model from xAI, released April 2026. Licence: proprietary.
Catalogue identity notes
Comparison-only (retired). xAI's first Speech to Text model, API slug grok-voice-transcribe-1.0. xAI release notes: April 2026 'Speech to Text is available ... generally available ... 25 languages with batch and streaming modes'; September 2026 'grok-voice-transcribe-2.0 is now available. Use grok-voice-transcribe-1.0 or grok-voice-transcribe-2.0'; October 2026 'grok-voice-transcribe-1.0 is deprecated and reaches end of life on October 2, 2026. All requests to that slug are routed to grok-voice-transcribe-2.0'. released = month the STT API (whose only model then was 1.0) went GA; the April note itself names no slug. Kept for its independent results run before end of life: verliapp/stt-benchmark @bfb5142 engine 'grok-stt' (xAI /v1/stt with no model field, runs 2026-07-16..18, when grok-voice-transcribe-1.0 was the only xAI STT model; 2.0 arrived in September 2026). raven.eval BENCHMARKS.md @f9f621e also has 'xai/grok-voice-transcribe-1.0' rows, not added because raven-de-read4-wer would still have fewer than 3 models. Successor: grok-voice-transcribe-2.
Score within the Open ASR English short-form WER (March 2026) comparison set only.
$ how it compares
The speech-to-text models ranked either side of Grok Voice Transcribe 1.0, out of 176 with a score, then more from xAI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 36 |  Tencent Hy-ASR-3.0-preview | 60±4 | comparison only | ||
| 37 |  Parakeet Unified EN 0.6B | 60±4 | local | ||
| 38 | Grok Voice Transcribe 1.0 | 60±5 | comparison only | ||
| 39 |  Qwen3-ASR-1.7B | 60±1 | api $0.0005/audio min +local | ||
| 40 | Canary-1B-Flash | 60±2 | local | ||
| more from xAI | |||||
| 2 | Grok Voice Transcribe 2.0 | 70±3 | api $0.0017/audio min | ||
see all speech-to-text models · find the ones that fit your machine
$ what it's good at
other tests4 not in a task · 91% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| verli STT benchmark · LibriSpeech test-clean WER | 1.8% ✎ | ▃·····▃▅█▃▅▅ beats 7/11 | 48% |
| verli STT benchmark · SPGISpeech 1,200-clip sample WER | 3.8% ✎ | ▂·····▄▃█▂▂▄ beats 3/11 | 22% |
| verli STT benchmark · Earnings-22 WER (2,741 utt.) | 9.5% ✎ | ▃·▃▅··▃▃▅▅█▃ beats 13/13 | 21% |
| raven.eval · German read/parliament 4-set WER mean (100 utt. each) | 8.4% ✎ | █··········█ beats 0/1 | — |
$ where to run it
No verified API or download recorded yet · add one
$ how the score was worked out · verliapp/stt-benchmark · inde… 91%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 verliapp/stt-benchmark · independent STT accuracy benchmark | 3 | 91% |
| prior · pull toward the middle | 9% | |
| weighted average | = 60.3 |
not counted (1)
| raven.eval · German ASR WER (public datasets) | raven.eval · German read/parliament 4-set WER mean (100 utt. each) [correct] | 8.4% | fewer than three models have this test yet, or this source ran it for this model alone |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json