Voice AI BenchmarksBy

OPENAISPEECH-TO-TEXT117,395 SAMPLES / 30 DAYSLAST RUN AUG 28, 2026, 21:30 UTC

official resource

GPT-4o mini Transcribe speech-to-text benchmarks

GPT-4o mini Transcribe is OpenAI's smaller GPT-4o-based transcription model.

Time to Final Segment#20 / 24
623ms
Word Error Rate#10 / 28
5.1%

Overview

OpenAI offers it as a dedicated Audio API transcription model with streaming output and prompt context for vocabulary guidance.

GPT-4o mini Transcribe is tested every day on fixed public audio — clean, accented, noisy, reverberant, far-field, clipped and phone-codec speech — for transcription accuracy and streaming latency.

Technical specifications

Made by
OpenAI
Hosted by
OpenAI
Source
Official API
Licensing
Proprietary
Deployment
Cloud
Region
US
Features
Multilingual, VAD, Keyterm biasing

How GPT-4o mini Transcribe ranks

Full STT dashboard
Time to Final Segment — every measured STT modelMilliseconds · lower is better · 30-day averageEvery STT model ranked on Time to Final Segment, with GPT-4o mini Transcribe highlighted.
  1. #1STT RT v564 ms
  2. #2Parakeet TDT 0.6B v370 ms
  3. #3STT 183 ms
  4. #4Nova 399 ms
  5. #5Nova 2101 ms
  6. #6Ink 2108 ms
  7. #7Scribe v2 Realtime120 ms
  8. #8Universal 3.5 Pro146 ms
  9. #9Defaultvia Azure155 ms
  10. #10Whisper Large v3180 ms
  11. #11Velma 2 STT Streaming191 ms
  12. #12Grok STT199 ms
  13. #13Pulse205 ms
  14. #14Defaultvia Speechmatics223 ms
  15. #15Defaultvia Gradium249 ms
  16. #16resonant-1288 ms
  17. #17Enhanced341 ms
  18. #18Voxtral Mini Transcribe Realtime 2602356 ms
  19. #19GPT Realtime Whisper559 ms
  20. #20GPT-4o mini Transcribe623 ms
Show all 24 models
  1. #21Solaria 1680 ms
  2. #22GPT-4o Transcribe742 ms
  3. #23Chirp 2811 ms
  4. #24Chirp 3813 ms
median of all models · 202 ms
STT models over the last 30 days, ranked on Time to Final Segment.
#ModelHostTTFSWERTTFTSamples
1STT RT v5Soniox64 ms1533 ms29,322
2Parakeet TDT 0.6B v3Together AI70 ms1210 ms29,161
3STT 1Inworld AI83 ms1468 ms29,280
4Nova 3Deepgram99 ms1434 ms29,290
5Nova 2Deepgram101 ms1436 ms29,145
6Ink 2Cartesia108 ms1812 ms29,288
7Scribe v2 RealtimeElevenLabs120 ms2157 ms29,313
8Universal 3.5 ProAssemblyAI146 ms1030 ms29,280
9DefaultAzure155 ms1792 ms6,682
10Whisper Large v3Together AI180 ms1241 ms29,171
11Velma 2 STT StreamingModulate191 ms1576 ms17,895
12Grok STTxAI199 ms—29,322
13PulseSmallest205 ms2072 ms29,313
14DefaultSpeechmatics223 ms1473 ms29,323
15DefaultGradium249 ms1979 ms28,235
16resonant-1Reson8288 ms—17,066
17EnhancedSpeechmatics341 ms1536 ms29,322
18Voxtral Mini Transcribe Realtime 2602Mistral356 ms1828 ms29,175
19GPT Realtime WhisperOpenAI559 ms1823 ms29,286
20GPT-4o mini TranscribeOpenAI623 ms—29,318
21Solaria 1Gladia680 ms1703 ms26,356
22GPT-4o TranscribeOpenAI742 ms—29,318
23Chirp 2Google811 ms6000 ms29,237
24Chirp 3Google813 ms5999 ms29,319
—FluxDeepgram—1089 ms29,354
—Flux MultilingualDeepgram—1175 ms29,344
—Nemotron 3.5 ASR StreamingTogether AI—1540 ms29,193
—Universal StreamingAssemblyAI—1510 ms29,297
Under-sampled models are excluded; tied models share a place. Dotted WER values split into substitutions, deletions and insertions on hover or tap.

Highest relative placement: 10th of 28 on Word Error Rate.

Latency vs accuracy

TTFS vs WEREach point is one measured model · GPT-4o mini Transcribe highlighted · 30-day averagesTime to Final Segment against Word Error Rate for every measured STT model, with GPT-4o mini Transcribe highlighted.
STT RT v5 — 64 ms, 5.8%Parakeet TDT 0.6B v3 — 70 ms, 11.0%STT 1 — 83 ms, 4.8%Nova 3 — 99 ms, 6.3%Nova 2 — 101 ms, 7.9%Ink 2 — 108 ms, 4.9%Scribe v2 Realtime — 120 ms, 5.5%Universal 3.5 Pro — 146 ms, 3.2%Default via Azure — 155 ms, 5.3%Whisper Large v3 — 180 ms, 8.2%Velma 2 STT Streaming — 191 ms, 6.4%Grok STT — 199 ms, 4.8%Pulse — 205 ms, 5.3%Default via Speechmatics — 223 ms, 5.5%Default via Gradium — 249 ms, 10.0%resonant-1 — 288 ms, 3.4%Enhanced — 341 ms, 4.3%Voxtral Mini Transcribe Realtime 2602 — 356 ms, 6.2%GPT Realtime Whisper — 559 ms, 5.2%Solaria 1 — 680 ms, 7.1%GPT-4o Transcribe — 742 ms, 4.7%Chirp 2 — 811 ms, 5.0%Chirp 3 — 813 ms, 4.1%GPT-4o mini Transcribe — 623 ms, 5.1%

Where the errors come from

WER compositionGPT-4o mini Transcribe's Word Error Rate split by error type · 30-day averageGPT-4o mini Transcribe's WER split into substitutions, deletions and insertions.
  • GPT-4o mini Transcribe5.1%
SubstitutionsDeletionsInsertions

Averages and tail latency

Averages hide slow outliers — these are the distributions behind each figure.

GPT-4o mini Transcribe latency distributionMilliseconds · shared axis across metrics · last 30 daysGPT-4o mini Transcribe's latency percentiles per metric: p25–p75 band, p50 tick, whisker to p99.
  • Time to Final Segmentp50 578 ms · p99 1371 ms
band p25–p75 · tick p50 · whisker to p99 with p90 and p95 stops
Average and percentile values per metric, with the number of samples behind each row.
MetricAveragep25p50p75p90p95p99Samples
Time to Final Segment623 ms507 ms578 ms676 ms783 ms934 ms1371 ms29,318
Word Error Rate5.1%0.0%0.0%6.7%13.6%21.1%50.0%29,359

Last 30 days

Daily medians from the same measurement runs · gaps are days without qualifying runs.

Time to Final Segment — daily p50Line p50 · band p25–p75 · UTC daysGPT-4o mini Transcribe's daily median Time to Final Segment over the last 30 days.
GPT-4o mini Transcribe · Jul 30: 557 msGPT-4o mini Transcribe · Jul 31: 534 msGPT-4o mini Transcribe · Aug 1: 505 msGPT-4o mini Transcribe · Aug 2: 506 msGPT-4o mini Transcribe · Aug 3: 590 msGPT-4o mini Transcribe · Aug 4: 569 msGPT-4o mini Transcribe · Aug 5: 602 msGPT-4o mini Transcribe · Aug 6: 541 msGPT-4o mini Transcribe · Aug 7: 554 msGPT-4o mini Transcribe · Aug 8: 568 msGPT-4o mini Transcribe · Aug 9: 549 msGPT-4o mini Transcribe · Aug 10: 549 msGPT-4o mini Transcribe · Aug 11: 528 msGPT-4o mini Transcribe · Aug 12: 573 msGPT-4o mini Transcribe · Aug 13: 510 msGPT-4o mini Transcribe · Aug 14: 548 msGPT-4o mini Transcribe · Aug 15: 521 msGPT-4o mini Transcribe · Aug 16: 554 msGPT-4o mini Transcribe · Aug 17: 562 msGPT-4o mini Transcribe · Aug 18: 513 msGPT-4o mini Transcribe · Aug 19: 555 msGPT-4o mini Transcribe · Aug 20: 708 msGPT-4o mini Transcribe · Aug 21: 658 msGPT-4o mini Transcribe · Aug 22: 556 msGPT-4o mini Transcribe · Aug 23: 599 msGPT-4o mini Transcribe · Aug 24: 643 msGPT-4o mini Transcribe · Aug 25: 699 msGPT-4o mini Transcribe · Aug 26: 670 msGPT-4o mini Transcribe · Aug 27: 671 msGPT-4o mini Transcribe · Aug 28: 647 ms
Word Error Rate — daily averageDaily average · UTC daysGPT-4o mini Transcribe's daily Word Error Rate over the last 30 days.
GPT-4o mini Transcribe · Jul 30: 4.9%GPT-4o mini Transcribe · Jul 31: 6.6%GPT-4o mini Transcribe · Aug 1: 6.7%GPT-4o mini Transcribe · Aug 2: 6.5%GPT-4o mini Transcribe · Aug 3: 5.1%GPT-4o mini Transcribe · Aug 4: 6.1%GPT-4o mini Transcribe · Aug 5: 6.6%GPT-4o mini Transcribe · Aug 6: 4.7%GPT-4o mini Transcribe · Aug 7: 6.5%GPT-4o mini Transcribe · Aug 8: 7.6%GPT-4o mini Transcribe · Aug 9: 6.2%GPT-4o mini Transcribe · Aug 10: 6.7%GPT-4o mini Transcribe · Aug 11: 6.3%GPT-4o mini Transcribe · Aug 12: 5.6%GPT-4o mini Transcribe · Aug 13: 6.5%GPT-4o mini Transcribe · Aug 14: 6.7%GPT-4o mini Transcribe · Aug 15: 5.8%GPT-4o mini Transcribe · Aug 16: 5.6%GPT-4o mini Transcribe · Aug 17: 6.0%GPT-4o mini Transcribe · Aug 18: 6.9%GPT-4o mini Transcribe · Aug 19: 6.3%GPT-4o mini Transcribe · Aug 20: 5.9%GPT-4o mini Transcribe · Aug 21: 4.8%GPT-4o mini Transcribe · Aug 22: 7.6%GPT-4o mini Transcribe · Aug 23: 5.4%GPT-4o mini Transcribe · Aug 24: 5.4%GPT-4o mini Transcribe · Aug 25: 5.3%GPT-4o mini Transcribe · Aug 26: 5.5%GPT-4o mini Transcribe · Aug 27: 5.3%GPT-4o mini Transcribe · Aug 28: 6.9%

Time to Final Segment by dataset

GPT-4o mini Transcribe by test conditionMilliseconds · lower is better · best condition firstGPT-4o mini Transcribe's Time to Final Segment on each benchmark dataset.
  1. WildASR accents557 ms
  2. WildASR phone codec613 ms
  3. WildASR reverb618 ms
  4. WildASR clean622 ms
  5. PipeCat (production)628 ms
  6. WildASR clipping628 ms
  7. WildASR noise gaps630 ms
  8. WildASR far-field633 ms
Time to Final Segment per dataset, with the number of samples behind each figure.
DatasetTTFSSamples
LibriSpeech883 ms1
ProductionPipeCat628 ms14,935
AccentsWildASR557 ms1,438
CleanWildASR622 ms5,779
ClippingWildASR628 ms1,444
Far-fieldWildASR633 ms1,442
Noise gapsWildASR630 ms1,441
Phone codecWildASR613 ms1,440
ReverbWildASR618 ms1,398

Strongest condition: WildASR accents at 557 ms · weakest: LibriSpeech at 883 ms.

Limits of this comparison

The mini model receives the same audio as GPT-4o Transcribe for a direct accuracy and latency comparison.

  • The results apply to this API model and Coval configuration, not every batch or Realtime transcription workflow.

Official sources

Results are re-measured daily using fixed datasets and reported over a rolling 30-day window. Same datasets, prompts and metric definitions for every model, measured by Coval’s open-source runner. Full methodology on the overview.

Evaluate your own voice agent

Use Coval to test your production configuration, prompts and calls—not only the public benchmark endpoints.

Book a Demo