OpenAI: Whisper Large V3
Whisper Large V3 is OpenAI's open-source automatic speech recognition model offering both audio transcription and translation. It supports 99+ languages and accepts common audio formats including mp3…
openai/whisper-large-v3
Whisper Large V3 is OpenAI's open-source automatic speech recognition model offering both audio transcription and translation. It supports 99+ languages and accepts common audio formats including mp3, mp4, wav, webm, flac, and ogg. With 1,550M parameters, it achieves a 10.3% word error rate and is well-suited for noise-robust, multilingual transcription in demanding conditions. Supports timestamp granularities at word and segment levels.
Specifications & capabilities
- Input
- audio
- Output
- transcription
- Context window · tokens
- 0
- Published model ID
- whisper-large-v3
- Capabilities
- audioInput, transcription
Description source ↗Evaluations
21–40 / 129
4.36%
- source id
- openASR
- leaderboard name
- 🇳🇱 Dutch · FLEURS (WER)
- dimension
- speech
- metric
- 🇳🇱 Dutch · FLEURS (WER)
- rank
- 11
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- multilingual/dutch
- suite
- OpenASR official public table
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- 🇳🇱 Dutch · FLEURS (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/multilingual/dutch.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 41
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗351.92
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 12
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_pt.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_pt.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 27
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗300.76
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 12
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_nl.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_nl.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 24
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗13.1%
- source id
- openASR
- leaderboard name
- Long-form · Average WER
- dimension
- speech
- metric
- Long-form · Average WER
- rank
- 12
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- longform
- suite
- OpenASR official public table
- datasets
- Earnings21
- Earnings22
- CORAAL
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- Long-form · Average WER
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/longform.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 14
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗4.11%
- source id
- openASR
- leaderboard name
- 🇳🇱 Dutch · MCV (WER)
- dimension
- speech
- metric
- 🇳🇱 Dutch · MCV (WER)
- rank
- 12
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- multilingual/dutch
- suite
- OpenASR official public table
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- 🇳🇱 Dutch · MCV (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/multilingual/dutch.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 41
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗3.92%
- source id
- openASR
- leaderboard name
- fr_mls (WER)
- dimension
- speech
- metric
- fr_mls (WER)
- rank
- 12
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_latest.csv
- metric_method
- published WER/RTFx
- metric_kind
- fr_mls (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_latest.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 34
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗3.15%
- source id
- openASR
- leaderboard name
- tedlium (WER)
- dimension
- speech
- metric
- tedlium (WER)
- rank
- 12
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/leaderboard_longform/longform_latest.csv
- metric_method
- published WER/RTFx
- metric_kind
- tedlium (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/leaderboard_longform/longform_latest.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 15
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗2.3%
- source id
- openASR
- leaderboard name
- 🇪🇸 Spanish · FLEURS (WER)
- dimension
- speech
- metric
- 🇪🇸 Spanish · FLEURS (WER)
- rank
- 12
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- multilingual/spanish
- suite
- OpenASR official public table
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- 🇪🇸 Spanish · FLEURS (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/multilingual/spanish.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 45
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗368.86
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 13
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_it.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_it.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 26
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗17.52%
- source id
- openASR
- leaderboard name
- Long-form · CORAAL (WER)
- dimension
- speech
- metric
- Long-form · CORAAL (WER)
- rank
- 13
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- longform
- suite
- OpenASR official public table
- datasets
- Earnings21
- Earnings22
- CORAAL
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- Long-form · CORAAL (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/longform.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 14
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗11.228%
- source id
- openASR
- leaderboard name
- Average (WER)
- dimension
- speech
- metric
- Average (WER)
- rank
- 13
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/leaderboard_longform/longform_latest.csv
- metric_method
- published WER/RTFx
- metric_kind
- Average (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/leaderboard_longform/longform_latest.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 16
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗4.1%
- source id
- openASR
- leaderboard name
- MCV WER (WER)
- dimension
- speech
- metric
- MCV WER (WER)
- rank
- 13
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_nl.csv
- metric_method
- published WER/RTFx
- metric_kind
- MCV WER (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_nl.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 36
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗386.26
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 14
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_es.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_es.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 27
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗339.65
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 14
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_fr.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_fr.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 27
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗327.55
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 14
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_de.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_de.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 27
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗4.81%
- source id
- openASR
- leaderboard name
- Avg (WER)
- dimension
- speech
- metric
- Avg (WER)
- rank
- 14
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_latest.csv
- metric_method
- published WER/RTFx
- metric_kind
- Avg (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_latest.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 34
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗3.5%
- source id
- openASR
- leaderboard name
- FLEURS WER (WER)
- dimension
- speech
- metric
- FLEURS WER (WER)
- rank
- 14
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_pt.csv
- metric_method
- published WER/RTFx
- metric_kind
- FLEURS WER (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_pt.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 40
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗3.2%
- source id
- openASR
- leaderboard name
- FLEURS WER (WER)
- dimension
- speech
- metric
- FLEURS WER (WER)
- rank
- 14
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_de.csv
- metric_method
- published WER/RTFx
- metric_kind
- FLEURS WER (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_de.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 41
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗110.92
- source id
- openASR
- leaderboard name
- RTFx
- dimension
- runtime
- metric
- RTFx
- rank
- 15
- aggregation scope
- status
- published
- direction
- higherIsBetter
- conditions
- suite
- hf-audio/multilingual_evals/multilingual_latest.csv
- metric_method
- published WER/RTFx
- metric_kind
- RTFx
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- hf-audio/multilingual_evals/multilingual_latest.csv
- published_at
- 2026-10-21T14:20:31Z
- source_revision
- 943867e1b478153a2161b0522c130917d0d96b01a780660e00252b573bc2cf13
- metric_population
- 27
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3
Original result ↗43.22%
- source id
- openASR
- leaderboard name
- 🇦🇲 Armenian · FLEURS (WER)
- dimension
- speech
- metric
- 🇦🇲 Armenian · FLEURS (WER)
- rank
- 15
- aggregation scope
- status
- published
- direction
- lowerIsBetter
- conditions
- view
- multilingual/armenian
- suite
- OpenASR official public table
- metric_method
- publisher WER/RTFx for selected public view
- space_revision
- 3b5317f4e49b6cb40502b95065b29fbfe08284ce
- publisher_version
- 13-11-2026
- metric_kind
- 🇦🇲 Armenian · FLEURS (WER)
- rank_method
- competition rank among reported evaluations in identical conditions, including unlinked observations
- source_file
- public/multilingual/armenian.json
- published_at
- 2026-11-13T11:42:34Z
- source_revision
- 8f2cc9c9f3b01b8670c8640efc67f24a3ea076c2bf5e90e7c674a1a829d69bd7
- metric_population
- 17
- benchmark_evaluation
- true
- source model id
- openai/whisper-large-v3 🤗
Original result ↗Published prices
1–1 / 1
$0.000008
- Inference provider
- openRouter
- Platform
- OpenRouter
- Unit
- second
- Currency
- USD
- Billing conditions
- direction
- input
- Billing label
- Input audio second
Provider specifications & access
1–11 / 11
- source model id
- HuggingFace/openai-whisper-large-v3@1
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@2
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@3
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@4
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@5
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@6
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@7
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- HuggingFace/openai-whisper-large-v3@8
- hosting
- foundry-managed-compute
- license
- apache-2.0
- source model id
- openai/whisper-large-v3
- hosting
- nvidia-catalog
- source model id
- publishers/openai/models/whisper-large@whisper-large-v3
- hosting
- vertex-self-deployed
- source model id
- whisper-large-v3
- hosting
- gateway
- max output tokens
- 1024
Safety assessments
1–1 / 1
35 / 100
- source id
- stanfordFMTI
- canonical model id
- openai/whisper-large-v3
- taxonomy
- transparency
- hazard category
- transparency
- hazard name
- Organization transparency
- policy area
- OpenAI
- assessment type
- organization-level disclosure
- metric
- disclosure_score
- result label
- 35 / 100
- sample size
- 100
- updated at
- 2026-11-09T18:08:05.227Z
Hugging Face model card
1–1 / 1
- downloads
- 3298898
- likes
- 6817
- pipeline tag
- automatic-speech-recognition
- library name
- transformers
- License
- apache-2.0
Explore more models