Model · OpenAI · Speech-To-Text
Whisper large-v3
High-accuracy multilingual speech-to-text. Best-in-class for non-English audio; the de-facto open baseline.
- Modality
- Speech-To-Text
- License
- MIT (Open)
- Parameter size
- 1.55B
- Released
- November 6, 2023
- Last verified
- June 8, 2026
- Runs locally
- Yes
What you need to run it
- Mac
- Yes. Any Apple Silicon Mac (M1 or newer) with at least 8 GB (16 GB recommended) of memory. Newer chips are faster, but memory is what decides whether it runs.
- Windows / Linux PC
- NVIDIA works best. A graphics card with at least 10 GB (12 GB recommended) of video memory.
- No graphics card
- Not practical. Too slow without a graphics card or Apple Silicon.
Runs well on Apple Silicon through whisper.cpp or MLX Whisper, but is slow on a CPU alone.
Numbers are for: official fp16 weights (1.55B); quantized whisper.cpp builds are smaller. Based on the maker’s own figures; where the maker gives only a graphics-card figure, the Mac memory is our estimate.
Strengths
- Multilingual (99 languages)
- Robust to noise and accents
- Open weights — runs offline
Weaknesses
- Slow on CPU; needs a GPU for real-time
- No speaker diarization built in
- Hallucinates on long silence
Try it
| Where | Type | Notes |
|---|---|---|
| Hugging Face | weights | Open weights, MIT |
| Replicate | hosted | API key required |
| Groq | hosted-api | Free tier |
| whisper.cpp | local | C/C++ port; runs on CPU and Apple Silicon |
Used in solutions
Version history
- Whisper large-v3 Turbo Oct 2024
- Whisper large-v3 Nov 2023 Current
- Whisper large-v2 Dec 2022 Deprecated
Official sources
- Model card model
- Whisper paper paper
- GitHub github
Change log
- — Added hardware requirements (Mac support, memory, NVIDIA). Source: https://github.com/openai/whisper (about 10 GB VRAM for large) + mlx-community/whisper-large-v3-mlx.
- — Initial entry.
Esc