Where audio is processed
Local models process audio on-device, while optional cloud models send it for cloud processing.
Sourced review · Verified August 24, 2026
Superwhisper review
Superwhisper review
local model choices
Superwhisper offers local dictation, a free small-model tier, and a choice of subscription or lifetime payment. Pro adds larger models and file transcription, while the vendor's offline performance guidance favors Apple Silicon.
WhisperJot is our product. Facts about Superwhisper come from SuperUltra, Inc.'s public pages and were last verified on the date shown.
This is a review of published product information, not a hands-on test or an independent security audit.
Best for: Local dictation with model choices
Strengths
What it does well
A free tier offers unlimited use of small AI models after the Pro trial, allowing ongoing dictation without a subscription.
Pro includes larger local models, optional cloud models, audio and video file transcription, and translation into English for broader transcription workflows.
At $84.99 annually, the subscription costs less than WhisperJot; monthly billing and a $249.99 lifetime license offer other payment choices.
Limitations
Where it falls short
Larger local models and file transcription require Pro, so the free tier's unlimited use does not include the full model selection.
The vendor says offline models run well on Apple Silicon and recommends cloud models for Intel Macs, qualifying its offline appeal.
There are macOS, Windows, and iOS apps, but no Linux or Android app; using your own API keys adds separate provider costs.
Data handling
Privacy check
Collection and training
The privacy policy says it does not collect audio, transcripts, personal information, or usage data, or train on user data.
Cloud policy scope
The policy does not separately explain data flows for optional cloud models or user-supplied API keys.
Fit
Who should pick it
Choose Superwhisper if small free models, Pro file transcription, or a lifetime license matter to you. Apple Silicon users seeking offline local models fit the vendor's performance guidance.
Our product
How WhisperJot compares
WhisperJot is private by default, with Jot Local and Jot Local Pro processing speech on-device. Local engines work fully offline after the first model download. WhisperJot Local ($5/month) and WhisperJot Pro ($99/year) both include unlimited local dictation, custom vocabulary, voice commands, and local history on stable Mac and Windows apps plus a Linux beta. Jot Cloud is our optional cloud engine on WhisperJot Pro: it is opt-in, runs only while selected, and audio is never stored on our servers. WhisperJot has no free tier; the browser demo offers a short dictation before purchase.
Questions
Superwhisper FAQ
Is Superwhisper private?
Superwhisper can process speech locally or through optional cloud models. Its policy says it collects no audio, transcripts, personal information, or usage data, does not train on user data, and retains no server data. However, it does not separately explain cloud-model or personal API-key flows; read the privacy policy with your selected model and provider in mind.
Is Superwhisper or WhisperJot cheaper?
Superwhisper's annual plan costs less than WhisperJot Pro: $84.99 per year versus $99 per year. Superwhisper also offers a free tier, an $8.49 monthly option, and a $249.99 lifetime license. WhisperJot Local is $5/month with unlimited on-device dictation; WhisperJot Pro is $99/year and adds optional Jot Cloud — opt-in before first use, used only while selected, and audio is never stored on our servers. WhisperJot has no free tier or lifetime option.
Does Superwhisper work fully offline?
Yes, Superwhisper works fully offline with local models on Apple Silicon. Its vendor says Intel Macs work best with cloud models and that offline models only run really well on Apple Silicon. WhisperJot's local engines work fully offline on its supported Apple Silicon Mac, Windows x64, and Linux x64 or arm64 builds after the first model download.
References
Sources
Product facts last verified August 24, 2026. Sources include the vendor pages and any supporting references listed below.