Platform edition
Voxtral for macOS
Voxtral locally through MLX on Apple silicon without cloud uploads.
- Version
- 1.0.14
- Plugin ID
- com.
typewhisper. voxtral - Requirements
- TypeWhisper 1.6.0 or newer · macOS 14.0 or newer
Inside this edition
Features in this edition
- Local speech-to-text processing
- Processing without uploading audio or text
- Model or runtime downloads managed by TypeWhisper
- MLX inference on Apple silicon
Settings in this edition
- Optionally store a Hugging Face token for more reliable model downloads
- Choose, download, and manage the local model
About
Voxtral brings Mistral’s speech recognition model to your Mac via Apple’s MLX framework. All processing runs locally with no cloud connection required. Voxtral supports 15 languages and offers real-time streaming transcription.
For Mistral’s cloud transcription with Voxtral Transcribe 3 and Voxtral Realtime, on Mac and Windows, use the Mistral AI add-on.
Features
- Mistral’s Voxtral model running locally via MLX
- 15 languages supported
- Real-time streaming transcription
- No API key or cloud account needed
Transcription Models
| Model | ID | Size |
|---|---|---|
| Voxtral Mini 4B 4-bit | voxtral-mini-4b-2602-4bit | ~2.5 GB |
| Voxtral Mini 4B FP16 | voxtral-mini-4b-2602-fp16 | ~8 GB |
Configuration
No API key required. Voxtral runs entirely on-device.
Setup
- In TypeWhisper, open Settings > Discover plugins and install Voxtral. The add-on then appears in the Settings sidebar, and its page opens.
- The model will be downloaded on first use.
- Select Voxtral as your transcription engine in Settings or in a profile.