Whisper Large v3
Sep 2022
Whisper Large v3Compare inon
Whisper Large v3 measured on 46 of 81 Speech to text benchmarks46
BenchmarkByScoreRank
No benchmark matches
$0.004per minute
10.1AA Speech to text
Sep 2022released
en, es, fr +86Languages
$0.004 / minPer audio minute
/ large-v3Price covers
Ranking?
Full ranking →···
52Gradium Speech-to-Text6.853Whisper Large v310.154Qwen3.5 Omni Flash13.5Strong at
Alternatives?
ModelScoreCostPriceWhat changes
CheaperMAI Transcribe 2×5.1-61%$0.002Languages: loses no, hr, la +31MBetterModulate STT Batch English VFast×2.4-90%$0Languages: not verified
OpenCanary Qwen 2.5B, NVIDIA×2.3-83%$0.001Languages: loses es, fr, de +85+1
CheaperVoxtral Small Transcribe 3×4.8-5%$0.004Languages: not verifiedCheaperVoxtral Mini Transcribe 3×4.6-17%$0.004Languages: not verifiedModelScoreCostPriceWhat changes
MAI Transcribe 2×5.1-61%$0.002Languages: loses no, hr, la +31Voxtral Small Transcribe 3×4.8-5%$0.004Languages: not verifiedVoxtral Mini Transcribe 3×4.6-17%$0.004Languages: not verified
Scribe v2×4.6-13%$0.004
Grok Voice Transcribe 2.0×4.4-61%$0.002Languages: loses he, ta, no +52ModelScoreCostPriceWhat changes
MModulate STT Batch English VFast×2.4-90%$0Languages: not verifiedWizper Large v3Open×2.1-88%$0.001Memory (4-bit): not verifiedWhisper Large v3 TurboOpen×2.2-84%$0.001Memory (4-bit): not verifiedCanary Qwen 2.5B, NVIDIAOpen×2.3-83%$0.001Languages: loses es, fr, de +85+1 Large v3Open+77%-65%$0.001Memory (4-bit): not verifiedModelScoreCostPriceWhat changes
Canary Qwen 2.5B, NVIDIAOpen×2.3-83%$0.001Languages: loses es, fr, de +85+1ModelScoreCostPriceWhat changes
Large v3Open+77%-65%$0.001Memory (4-bit): not verifiedWizper Large v3Open×2.1-88%$0.001Memory (4-bit): not verifiedWhisper Large v3 TurboOpen×2.2-84%$0.001Memory (4-bit): not verifiedGPT-4o Mini Transcribeunstated×2.2-29%$0.003Languages: not verifiedWhisper Large v2Open×2.5+42%$0.006Per audio minute: $0.004 / min → $0.006 / min+1GPT-4o Transcribe×2.5+42%$0.006Languages: not verified+1GPT Transcribe, OpenAI×3.1+6%$0.004Languages: loses de, ja, pt +82GPT Live Transcriben/a—$0.017Languages: not verified+1All OpenAI models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use
History
2 versionsBenchmarks
in Speech & VoiceOpen ASR · average WER7.8#44
Open ASR · RTFx470.7#32
Open ASR · AMI Cleaned WER13.6#52
Open ASR · URGENT 2024 WER13.7#35
Open ASR · AMI WER14.9#24
Open ASR · Earnings-22 WER11.6#32
Open ASR · VoxPopuli WER8.7#30
Open ASR · French RTFx339.7#13
Open ASR · German RTFx327.6#13
Open ASR · Spanish RTFx386.3#13
Open ASR · Italian RTFx368.9#12
Open ASR · Portuguese RTFx351.9#11
Open ASR · Hindi RTFx89.6#9
Open ASR · Hindi Monsoon WER28.2#28
Open ASR · Dutch RTFx300.8#11
Open ASR · Dutch MCV WER4.1#13
Open ASR · Armenian RTFx73.4#6
FAQ
How much does Whisper Large v3 cost?›
$0.004 per minute.
Is Whisper Large v3 free to use?›
There is no free API; it costs $0.004 per minute. The weights are open, so you can run it yourself.
How good is Whisper Large v3, and what is it best at?›
#49 of 51 in Speech to text.
Is Whisper Large v3 better than Gradium Speech-to-Text?›
On AA Speech to text · Recorded audio word error rate, Whisper Large v3 scores 10.1 vs 6.8 — lower. It is 67% cheaper.
When was Whisper Large v3 released? Is there a newer version?›
Released 2022-09-27. It is the current version in its line.
Is Whisper Large v3 open source?›
Yes, the weights are open (MIT licence).
Can I run Whisper Large v3 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is Whisper Large v3?›
Whisper Large v3 is an open-weights speech & voice model from OpenAI, released Sep 2022.
What are the best alternatives to Whisper Large v3?›
MAI Transcribe 2 (cheaper), Modulate STT Batch English VFast (better), Canary Qwen 2.5B, NVIDIA (open weights).