Whisper Large V3 Turbo
Q4 GGUF download: handy-computer/whisper-large-v3-turbo-gguf - 0.54 GB on disk, read 2026-09-27. VRAM figures above are estimates; this is the actual file size.
Will it run on your hardware?
Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest
Need an exact figure for your context length? Use the VRAM calculator.
Run it locally
Copy-paste - running in under a minute
vllm serve openai/whisper-large-v3-turboNew to this? Start with Ollama · serve to many users with vLLM.
Deep dive
Notes, sources, and the full write-up
Whisper Large V3 Turbo
Whisper Large V3 Turbo is OpenAI's state-of-the-art automatic speech recognition model. It's a pruned version of Whisper Large V3 with 4 decoder layers (vs 32), making it dramatically faster with minimal quality loss. 8.38M monthly downloads - the most popular ASR model.
Key features
- 99 languages - multilingual speech recognition
- 8.38M downloads/month - most popular ASR model
- 809M params - efficient size
- MIT license - fully open, commercial use OK
- Speech translation - transcribe to English from any language
- Torch compile - 4.5x speedup with torch.compile
Performance
| Benchmark | WER (%) |
|---|---|
| LibriSpeech Clean | 2.10 |
| LibriSpeech Other | 4.24 |
| Common Voice 15 | 7.83 |
| GigaSpeech | 10.14 |
| AMI | 16.13 |
| Earnings-22 | 11.63 |
| SPGISpeech | 2.97 |
Quick start
from transformers import pipeline
pipe = pipeline(
"automatic-speech-recognition",
model="openai/whisper-large-v3-turbo",
device="cuda:0"
)
result = pipe("audio.mp3")
print(result["text"])When to use
- Transcription - meetings, lectures, podcasts
- Subtitling - automatic subtitle generation
- Voice assistants - speech-to-text pipeline
- Language learning - speech translation
- Accessibility - real-time captioning
From the community
What users are saying
“Hi everyone! Thanks for waiting while we took a deeper look. We've just applied a change so that this can be deployed easily from the model card, and from our model catalog in Inference Endpoints here: https://endpoints.huggingface.co/catalog?query=whisper.”
“Understood, have you tried using a byte 64 encoded audio file? In the docs it uses tricky logic like, “If not using parameters you can also just use an audio file directly” so you may have to first convert the to byte 64 encoding. Let me know if I misunderstood, I spent all day on this 😂”
“have your problem been solved yet？ I met the same issue. By use original openai code I can set initial prompt but can not fintune.”
Frequently asked
Quick answers to common questions
How much VRAM does Whisper Large V3 Turbo need?
Whisper Large V3 Turbo with 0.8B parameters needs significant VRAM depending on quantization. Use our VRAM calculator for an exact estimate.
Is Whisper Large V3 Turbo better than other Whisper models?
Whisper Large V3 Turbo has 0.8B parameters with 8,192 context - a strong choice for speech-recognition, transcription, translation.
What license is Whisper Large V3 Turbo under?
Whisper Large V3 Turbo is released under the MIT license, making it suitable for most commercial and personal projects.
What hardware runs Whisper Large V3 Turbo well?
With 0.8B parameters, Whisper Large V3 Turbo requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.
What is the best quantization for Whisper Large V3 Turbo?
Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Step up to Q5_K_M or Q8_0 only if you have spare VRAM. Use our VRAM calculator to compare.
What models compete with Whisper Large V3 Turbo?
Whisper Large V3 Turbo competes with other models in its class. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.
Compare & pair with
Similar models and recommended hardware
Related models
Nearby options
Similar models and compatible hardware by spec