Project AURIONAI Use-Cases Research & Innovation
speech recognition

Whisper large-v3

Whisper large-v3; led the Open ASR Leaderboard in 2023.

OpenAIOpenAIUSAWhisperFamily
OpenAI
Levelvariant
Classspeech recognition
Parameters1.55B
Contextnot in the notes
Released2023
Opennessopen weights permissive
Runs onsingle datacentre GPU
Statusactive

Benchmark Results

maintainer-run
Hugging Face Open ASR Leaderboardspeech
leadership historymaintainer-run
7.44
048

Score in WER. Lower is better.

leader in 20232023 select.keenable.ai
average WER per Gladia, from the same leaderboardmaintainer-run
7.4
048

Score in WER. Lower is better.

gladia.io
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).

Lineage

finetunedistillsuccessor
Gemma2-2b-IT to MERaLiON-2-3B: finetune, medium confidenceGemma2-9b-IT to MERaLiON-2-10B: finetune, high confidenceWhisper large-v3 to MERaLiON-2-10B: finetune, medium confidenceWhisper large-v3 to MERaLiON-2-3B: finetune, medium confidenceWhisper large-v3 to distil-large-v3: distill, high confidenceWhisper large-v3 to distil-large-v3.5: distill, high confidencedistil-large-v3 to distil-large-v3.5: successor, high confidenceGemma2-9b-ITGemma2-9b-ITGoogleGemma2-2b-ITGemma2-2b-ITGoogleWhisper large-v3Whisper large-v3OpenAI, 2023distil-large-v3distil-large-v3Hugging Face, Mar 2024distil-large-v3.5distil-large-v3.5Hugging Face, Dec 2024MERaLiON-2-10BMERaLiON-2-10BA*STAR (Institute for Infocomm Re…MERaLiON-2-3BMERaLiON-2-3BA*STAR (Institute for Infocomm Re…
A*STAR (Institute for Infocomm Research, I2R)MERaLiON-2-10B is a child (finetune)medium confidencemodel card
A*STAR (Institute for Infocomm Research, I2R)MERaLiON-2-3B is a child (finetune)medium confidencemodel card
Hugging Facedistil-large-v3 is a child (distill)high confidencemodel card
Hugging Facedistil-large-v3.5 is a child (distill)high confidencemodel card
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 7 models. Confidence and evidence are the notes' own.

R&D Topics

Massively multilingual speech recognitionproduced by

Identity

Developer
OpenAI
Family
Whisper
Released
2023 sourcesnotesunverified
Weights published
–
Record checked
10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
autoregressive
Total parameters
1.55B sources
Active parameters
–
Checkpoint parameters
–
Context length
not in the notes
Input
audio
Output
text
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
open weights permissive
Licence
MIT License sources
Access routes
open weights download
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
single datacentre GPU
Footprint
not in the notes
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
99+ languages
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Industry Use Cases

No industry use case links this model.

Sources

  1. Gladia - Best open-source speech-to-text models in 2026 gladia.io https://www.gladia.io/blog/best-open-source-speech-to-text-models
  2. Speech Recognition Accuracy Leaders, 2022–2026 select.keenable.ai https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Whisper Large-v3", https://aurion.semai.id/models/openai-whisper-large-v3/, records of October 2026.