Downloading Your First Whisper Model
ScribeForge doesn't ship with a Whisper model baked in — you choose and download one the first time you use the app, via Settings → General → Default model, or the model picker in the Transcribe screen. After that single download, transcription never needs the internet again.
Which model should you pick?
Whisper comes in several sizes. Smaller models are faster and use less disk space; larger models are more accurate, especially on accents, background noise, and technical vocabulary.
- tiny / base — fastest, smallest download, good for quick drafts and clear audio.
- small — a solid middle ground for everyday use. These three (tiny, base, small) are the models available on the free plan.
- medium / large — most accurate, best for noisy audio, accents, or professional/legal use. Available on Pro plans.
If you're unsure, start with small — it covers most everyday transcription well without a long download.
Downloading a model
- Open Settings → General (or the model dropdown on the Transcribe screen).
- Pick a model from the list — each shows its approximate download size.
- Wait for the download to complete. It only happens once per model; ScribeForge stores it locally.
- Set it as your default model, or pick it per-transcription from the Transcribe screen.
Free vs. Pro models
The free plan includes the tiny, base and small models. Upgrading to a paid plan unlocks the larger, more accurate models — see pricing for details, or read more about the model itself on the Whisper transcription page.