Voice AI BenchmarksBy

GLADIASPEECH-TO-TEXT131,920 SAMPLES / 30 DAYSLAST RUN AUG 28, 2026, 21:30 UTC

official resource

Solaria 1 speech-to-text benchmarks

Solaria 1 is Gladia's real-time speech recognition model.

Time to Final Segment#21 / 24
680ms
Word Error Rate#22 / 28
7.1%
Time to First Token#15 / 24
1703ms

Overview

Gladia's service combines multilingual transcription with translation and code switching; Coval's endpoint also includes voice activity and keyterm controls.

Solaria 1 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
Gladia
Hosted by
Gladia
Source
Official API
Licensing
Proprietary
Deployment
Cloud
Region
Europe
Features
Multilingual, VAD, Translation, Code switching, Keyterm biasing

How Solaria 1 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 Solaria 1 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
  21. #21Solaria 1680 ms
Show all 24 models
  1. #22GPT-4o Transcribe742 ms
  2. #23Chirp 2811 ms
  3. #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: 15th of 24 on Time to First Token.

Latency vs accuracy

TTFS vs WEREach point is one measured model · Solaria 1 highlighted · 30-day averagesTime to Final Segment against Word Error Rate for every measured STT model, with Solaria 1 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%GPT-4o mini Transcribe — 623 ms, 5.1%GPT-4o Transcribe — 742 ms, 4.7%Chirp 2 — 811 ms, 5.0%Chirp 3 — 813 ms, 4.1%Solaria 1 — 680 ms, 7.1%

Where the errors come from

WER compositionSolaria 1's Word Error Rate split by error type · 30-day averageSolaria 1's WER split into substitutions, deletions and insertions.
  • Solaria 17.1%
SubstitutionsDeletionsInsertions

Averages and tail latency

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

Solaria 1 latency distributionMilliseconds · shared axis across metrics · last 30 daysSolaria 1's latency percentiles per metric: p25–p75 band, p50 tick, whisker to p99.
  • Time to Final Segmentp50 393 ms · p99 6796 ms
  • Time to First Tokenp50 1598 ms · p99 6630 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 Segment680 ms332 ms393 ms515 ms950 ms2208 ms6796 ms26,356
Word Error Rate7.1%0.0%3.3%10.0%18.2%25.9%53.3%26,391
Time to First Token1703 ms1198 ms1598 ms1798 ms2396 ms3185 ms6630 ms26,391

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 daysSolaria 1's daily median Time to Final Segment over the last 30 days.
Solaria 1 · Aug 1: 328 msSolaria 1 · Aug 2: 394 msSolaria 1 · Aug 3: 401 msSolaria 1 · Aug 4: 419 msSolaria 1 · Aug 5: 457 msSolaria 1 · Aug 6: 413 msSolaria 1 · Aug 7: 467 msSolaria 1 · Aug 8: 338 msSolaria 1 · Aug 9: 367 msSolaria 1 · Aug 10: 395 msSolaria 1 · Aug 11: 391 msSolaria 1 · Aug 12: 408 msSolaria 1 · Aug 13: 377 msSolaria 1 · Aug 14: 366 msSolaria 1 · Aug 15: 384 msSolaria 1 · Aug 16: 366 msSolaria 1 · Aug 17: 395 msSolaria 1 · Aug 18: 387 msSolaria 1 · Aug 19: 476 msSolaria 1 · Aug 20: 464 msSolaria 1 · Aug 21: 392 msSolaria 1 · Aug 22: 376 msSolaria 1 · Aug 23: 370 msSolaria 1 · Aug 24: 415 msSolaria 1 · Aug 25: 398 msSolaria 1 · Aug 26: 511 msSolaria 1 · Aug 27: 446 msSolaria 1 · Aug 28: 422 ms
Word Error Rate — daily averageDaily average · UTC daysSolaria 1's daily Word Error Rate over the last 30 days.
Solaria 1 · Aug 1: 6.1%Solaria 1 · Aug 2: 6.5%Solaria 1 · Aug 3: 6.6%Solaria 1 · Aug 4: 8.0%Solaria 1 · Aug 5: 8.3%Solaria 1 · Aug 6: 7.2%Solaria 1 · Aug 7: 8.8%Solaria 1 · Aug 8: 8.1%Solaria 1 · Aug 9: 7.7%Solaria 1 · Aug 10: 8.0%Solaria 1 · Aug 11: 9.9%Solaria 1 · Aug 12: 7.7%Solaria 1 · Aug 13: 7.4%Solaria 1 · Aug 14: 8.1%Solaria 1 · Aug 15: 7.0%Solaria 1 · Aug 16: 8.6%Solaria 1 · Aug 17: 7.1%Solaria 1 · Aug 18: 9.5%Solaria 1 · Aug 19: 8.8%Solaria 1 · Aug 20: 7.4%Solaria 1 · Aug 21: 8.1%Solaria 1 · Aug 22: 8.3%Solaria 1 · Aug 23: 7.6%Solaria 1 · Aug 24: 7.5%Solaria 1 · Aug 25: 7.0%Solaria 1 · Aug 26: 7.7%Solaria 1 · Aug 27: 7.1%Solaria 1 · Aug 28: 7.9%

Time to Final Segment by dataset

Solaria 1 by test conditionMilliseconds · lower is better · best condition firstSolaria 1's Time to Final Segment on each benchmark dataset.
  1. PipeCat (production)647 ms
  2. WildASR reverb670 ms
  3. WildASR clean704 ms
  4. WildASR noise gaps719 ms
  5. WildASR far-field726 ms
  6. WildASR clipping733 ms
  7. WildASR accents735 ms
  8. WildASR phone codec748 ms
Time to Final Segment per dataset, with the number of samples behind each figure.
DatasetTTFSSamples
LibriSpeech260 ms1
ProductionPipeCat647 ms13,723
AccentsWildASR735 ms1,300
CleanWildASR704 ms5,186
ClippingWildASR733 ms1,246
Far-fieldWildASR726 ms1,233
Noise gapsWildASR719 ms1,233
Phone codecWildASR748 ms1,240
ReverbWildASR670 ms1,194

Strongest condition: LibriSpeech at 260 ms · weakest: WildASR phone codec at 748 ms.

Limits of this comparison

Coval evaluates the official API on fixed inputs, including difficult acoustic conditions a clean-only score would hide.

  • Coval scores transcription accuracy and timing, not translation quality or a complete language matrix.

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