Three-way feature table
Vendor plans and feature lists change. Confirm the current details on each vendor's page before paying.
| Feature | Superwhisper | MacWhisper | StarWhisper |
| Platforms | Mac, Windows, iOS, Android | macOS | Windows, macOS (Apple Silicon) |
| Offline capability | Local Whisper processing | Local Whisper processing | Local Whisper via whisper.cpp |
| Pricing model | Free tier and paid subscription, check vendor | Free version and one-time Pro options have been offered, check vendor | Free tier, then $10/month or $80/year Pro |
| Free tier | Free tier listed, check current limits | Free version listed, check current limits | 2,000/week |
| Languages | Whisper language support, check current vendor list | Whisper language support, check current vendor list | 99 languages |
| Hotkey and system-wide insertion | Hotkey-based live dictation into Mac apps | File-first workflow, app insertion varies by feature | Global hotkey types into the active app |
| Local acceleration | Mac hardware path | Mac hardware path | CUDA, Vulkan, and Metal support |
Same model family, different jobs
It is easy to treat three Whisper apps as interchangeable. The more useful question is where the text should come from and where it should go. MacWhisper is associated with taking an existing recording and turning it into a transcript. That suits interviews, lectures, and other files you want to review later.
Superwhisper is associated with a live Mac dictation loop. Focus a text field, use the app's hotkey, speak, and continue writing without opening a file-transcription workspace. If that is how you think and you use a Mac all day, its focus is relevant.
StarWhisper is our product and focuses on the same live text loop across Windows and macOS. A global hotkey starts dictation, local Whisper processing creates text, and the result is inserted into the active app. The choice is not a claim that one workflow is best for everyone. It is a choice about where you spend your time.
Privacy, offline use, and hardware
The comparison material describes these apps as local, offline-first Whisper options. A local model can process speech without a transcription upload, which is useful when the network is unavailable or when your work requires a local data path. Check current privacy statements and optional online features before using any app for sensitive work.
StarWhisper supports model sizes from tiny through large-v3 and supports CUDA, Vulkan, and Metal acceleration where available. That gives you a way to balance local workload and model size. Hardware still matters, so test the model and microphone on the computer where you will dictate.
Pricing and a fair test
StarWhisper's free tier includes 2,000 words per week. Pro is $10 per month or $80 per year. MacWhisper and Superwhisper have their own free and paid plans, and current amounts should be checked with the vendors instead of copied from an old page.
Use the same short passage, microphone, and destination app for each test. Check live insertion, correction effort, offline behavior, and any file workflow you need. For more detail, read StarWhisper vs Superwhisper, StarWhisper vs MacWhisper, and the approved StarWhisper facts.