Voice AI BenchmarksBy

ALIBABATEXT-TO-SPEECH11,459 SAMPLES / 30 DAYSLAST RUN NOV 2, 2026, 09:00 UTC

Qwen3 TTS Fast text-to-speech benchmarks

Qwen3 TTS Fast, hosted by Nari Labs, measures mean 68 ms time to first audio (3rd of 32) and 1.8% word error rate (18th of 32) among TTS systems. Results cover the last 30 days. Coval benchmarks hosted endpoints daily using a consistent evaluation methodology. Last measured .

Coval has active text-to-speech measurements for Qwen3 TTS Fast, created by Alibaba and served on Nari Labs.

Time to First Audio#3 / 32
68ms
TTFA Network Roundtrip#2 / 32
58ms
TTFA Leading Silence#6 / 32
10ms
Word Error Rate#18 / 32
1.8%

Overview

Qwen3 TTS Fast is tested every day on a fixed set of text prompts, measuring how quickly audible speech starts and how intelligible the result is.

Technical specifications

Made by
Alibaba
Hosted by
Nari Labs
Source
Shared inference
Licensing
Open-weight
Deployment
Cloud
Region
US
Features
Multilingual

How Qwen3 TTS Fast ranks

Full TTS dashboard
Time to First Audio — every measured TTS modelMilliseconds · lower is better · 30-day averageEvery TTS model ranked on Time to First Audio, with Qwen3 TTS Fast highlighted.
  1. #1vui42 ms
  2. #2Default64 ms
  3. #3Qwen3 TTS Fast68 ms
  4. #4TTS Flash 283 ms
  5. #5Qwen3 TTS 1.7b105 ms
  6. #6Simba 3.0110 ms
  7. #7Simba 3.2118 ms
  8. #8Palabra TTS123 ms
  9. #9PhantomZ 4.0141 ms
  10. #10TTS 2163 ms
  11. #11drama3-preview165 ms
  12. #12Falcon 2177 ms
Show all 32 models
  1. #13Eleven v4 Turbo194 ms
  2. #14Flash v2.5202 ms
  3. #15Flux TTS207 ms
  4. #16Assemblyai TTS 1224 ms
  5. #17TTS RT v1261 ms
  6. #18TTS Rt v2264 ms
  7. #19Airy TTS281 ms
  8. #20Sonic 3.6287 ms
  9. #21Sonic 3.5291 ms
  10. #22Aura 2303 ms
  11. #23Grok TTS342 ms
  12. #24S2.1 Pro367 ms
  13. #25Coda378 ms
  14. #26Lightning v3.1 Pro383 ms
  15. #27Mist v3387 ms
  16. #28Daytona TTS439 ms
  17. #29Aura 2 En546 ms
  18. #30Chirp 3 HD588 ms
  19. #31Qwen3 TTS Flash Realtime727 ms
  20. #32S2.1 Pro Free802 ms
median of all models · 242 ms
TTS models over the last 30 days, ranked on Time to First Audio.
#ModelHostTTFAWERSamples
1vuiFluxions42 ms2,663
2Gradium TTS Beta 202609Gradium64 ms2,878
3Qwen3 TTS FastNari Labs68 ms2,878
4TTS Flash 2Inworld AI83 ms2,878
5Qwen3 TTS 1.7bBaseten105 ms2,875
6Simba 3.0SpeechifyAI110 ms2,875
7Simba 3.2SpeechifyAI118 ms2,875
8Palabra TTS v1Palabra123 ms2,351
9Phantom Z 3.4 conversationalDeepdub141 ms2,866
10TTS 2Inworld AI163 ms2,876
11Drama 3 Preview Us EastFish Audio165 ms1,955
12Falcon 2Murf177 ms2,875
13Eleven v4 TurboElevenLabs194 ms2,878
14Flash v2.5ElevenLabs202 ms2,869
15Flux TTSDeepgram207 ms2,754
16Assemblyai TTS 1AssemblyAI224 ms727
17TTS RT v1Soniox261 ms2,418
18TTS Rt v2Soniox264 ms2,405
19Airy TTSOut of Set281 ms2,177
20Sonic 3.6Cartesia287 ms2,877
21Sonic 3.5Cartesia291 ms2,877
22Aura 2Deepgram303 ms2,878
23Grok TTSSpaceXAI342 ms2,876
24S2.1 ProFish Audio367 ms2,873
25CodaRime378 ms506
26Lightning v3.1 ProSmallest383 ms2,867
27Mist v3Rime387 ms506
28Daytona TTSGuava439 ms2,880
29Aura 2 EnCloudflare546 ms2,875
30Chirp 3 HDGoogle588 ms2,878
31Qwen3 TTS Flash RealtimeAlibaba727 ms2,877
32S2.1 Pro FreeFish Audio802 ms2,860
Under-sampled models are excluded; tied models share a place. Dotted WER values split into substitutions, deletions and insertions on hover or tap.

Latency vs accuracy

TTFA vs WEREach point is one measured model · Qwen3 TTS Fast highlighted · 30-day averagesTime to First Audio against Word Error Rate for every measured TTS model, with Qwen3 TTS Fast highlighted.
vui — 42 ms, 2.0%Default — 64 ms, 1.9%TTS Flash 2 — 83 ms, 1.7%Qwen3 TTS 1.7b — 105 ms, 1.9%Simba 3.0 — 110 ms, 1.6%Simba 3.2 — 118 ms, 2.0%Palabra TTS — 123 ms, 2.7%PhantomZ 4.0 — 141 ms, 2.2%TTS 2 — 163 ms, 1.6%drama3-preview — 165 ms, 1.6%Falcon 2 — 177 ms, 1.6%Eleven v4 Turbo — 194 ms, 1.7%Flash v2.5 — 202 ms, 1.9%Flux TTS — 207 ms, 1.8%Assemblyai TTS 1 — 224 ms, 1.6%TTS RT v1 — 261 ms, 2.6%TTS Rt v2 — 264 ms, 2.7%Airy TTS — 281 ms, 1.6%Sonic 3.6 — 287 ms, 1.5%Sonic 3.5 — 291 ms, 1.5%Aura 2 — 303 ms, 1.9%Grok TTS — 342 ms, 1.7%S2.1 Pro — 367 ms, 1.6%Coda — 378 ms, 2.4%Lightning v3.1 Pro — 383 ms, 1.6%Mist v3 — 387 ms, 2.0%Daytona TTS — 439 ms, 1.8%Aura 2 En — 546 ms, 2.1%Chirp 3 HD — 588 ms, 1.6%Qwen3 TTS Flash Realtime — 727 ms, 3.0%S2.1 Pro Free — 802 ms, 1.6%Qwen3 TTS Fast — 68 ms, 1.8%

Where the errors come from

WER compositionQwen3 TTS Fast's Word Error Rate split by error type · 30-day averageQwen3 TTS Fast's WER split into substitutions, deletions and insertions.
  • Qwen3 TTS Fast1.8%
SubstitutionsDeletionsInsertions

Averages and tail latency

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

Qwen3 TTS Fast latency distributionMilliseconds · shared axis across metrics · last 30 daysQwen3 TTS Fast's latency percentiles per metric: p25–p75 band, p50 tick, whisker to max.
  • Time to First Audiop50 63 ms · Max 760 ms
  • TTFA Network Roundtripp50 53 ms · Max 750 ms
  • TTFA Leading Silencep50 10 ms · Max 10 ms
band p25–p75 · tick p50 · whisker to max with p90 and p95 stops
Average and percentile values per metric, with the number of samples behind each row.
MetricAveragep25p50p75p90p95MaxSamples
Time to First Audio68 ms61 ms63 ms69 ms81 ms99 ms760 ms2,878
TTFA Network Roundtrip58 ms51 ms53 ms59 ms71 ms89 ms750 ms2,878
TTFA Leading Silence10 ms10 ms10 ms10 ms10 ms10 ms10 ms2,878
Word Error Rate1.8%0.0%0.0%2.7%7.7%15.2%50.0%2,825

Last 30 days

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

Time to First Audio — daily p50Line p50 · band p25–p75 · UTC daysQwen3 TTS Fast's daily median Time to First Audio over the last 30 days.
Qwen3 TTS Fast · Oct 3: 65 msQwen3 TTS Fast · Oct 4: 59 msQwen3 TTS Fast · Oct 5: 87 msQwen3 TTS Fast · Oct 6: 69 msQwen3 TTS Fast · Oct 7: 64 msQwen3 TTS Fast · Oct 8: 75 msQwen3 TTS Fast · Oct 9: 70 msQwen3 TTS Fast · Oct 10: 65 msQwen3 TTS Fast · Oct 11: 69 msQwen3 TTS Fast · Oct 12: 70 msQwen3 TTS Fast · Oct 13: 69 msQwen3 TTS Fast · Oct 14: 66 msQwen3 TTS Fast · Oct 15: 66 msQwen3 TTS Fast · Oct 16: 65 msQwen3 TTS Fast · Oct 17: 60 msQwen3 TTS Fast · Oct 18: 88 msQwen3 TTS Fast · Oct 19: 60 msQwen3 TTS Fast · Oct 20: 68 msQwen3 TTS Fast · Oct 21: 62 msQwen3 TTS Fast · Oct 22: 68 msQwen3 TTS Fast · Oct 23: 74 msQwen3 TTS Fast · Oct 24: 65 msQwen3 TTS Fast · Oct 25: 67 msQwen3 TTS Fast · Oct 26: 60 msQwen3 TTS Fast · Oct 27: 88 msQwen3 TTS Fast · Oct 28: 76 msQwen3 TTS Fast · Oct 29: 61 msQwen3 TTS Fast · Oct 30: 67 msQwen3 TTS Fast · Oct 31: 78 msQwen3 TTS Fast · Nov 1: 70 msQwen3 TTS Fast · Nov 2: 61 ms
Word Error Rate — daily averageDaily average · UTC daysQwen3 TTS Fast's daily Word Error Rate over the last 30 days.
Qwen3 TTS Fast · Oct 3: 3.7%Qwen3 TTS Fast · Oct 4: 9.3%Qwen3 TTS Fast · Oct 5: 5.7%Qwen3 TTS Fast · Oct 6: 3.8%Qwen3 TTS Fast · Oct 7: 5.0%Qwen3 TTS Fast · Oct 8: 6.7%Qwen3 TTS Fast · Oct 9: 3.9%Qwen3 TTS Fast · Oct 10: 2.2%Qwen3 TTS Fast · Oct 11: 3.5%Qwen3 TTS Fast · Oct 12: 6.4%Qwen3 TTS Fast · Oct 13: 6.1%Qwen3 TTS Fast · Oct 14: 6.7%Qwen3 TTS Fast · Oct 15: 6.4%Qwen3 TTS Fast · Oct 16: 6.8%Qwen3 TTS Fast · Oct 17: 5.2%Qwen3 TTS Fast · Oct 18: 3.3%Qwen3 TTS Fast · Oct 19: 6.7%Qwen3 TTS Fast · Oct 20: 3.5%Qwen3 TTS Fast · Oct 21: 6.3%Qwen3 TTS Fast · Oct 22: 6.7%Qwen3 TTS Fast · Oct 23: 4.5%Qwen3 TTS Fast · Oct 24: 4.2%Qwen3 TTS Fast · Oct 25: 6.0%Qwen3 TTS Fast · Oct 26: 9.0%Qwen3 TTS Fast · Oct 27: 3.7%Qwen3 TTS Fast · Oct 28: 4.2%Qwen3 TTS Fast · Oct 29: 3.7%Qwen3 TTS Fast · Oct 30: 7.6%Qwen3 TTS Fast · Oct 31: 6.6%Qwen3 TTS Fast · Nov 1: 5.8%Qwen3 TTS Fast · Nov 2: 4.3%

How fast is Qwen3 TTS Fast?

On Nari Labs, Qwen3 TTS Fast measures mean 68 ms time to first audio (3rd of 32). Last measured 2026-11-02.

How accurate is Qwen3 TTS Fast?

On Nari Labs, Qwen3 TTS Fast measures 1.8% word error rate (18th of 32). Last measured 2026-11-02.

Who hosts Qwen3 TTS Fast?

Qwen3 TTS Fast is created by Alibaba and served by Nari Labs. Coval measures each hosted endpoint separately.

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