This is the hard one. Most dictation apps ship your voice to a server and call it magic; Superwhisper doesn't. It runs local models on your Mac, it's well-reviewed, it's been around, and it has a permanent free tier that includes unlimited use of its small local models and voice-to-text in any app.
So let's start where an honest page should start: if Superwhisper's free tier covers you, use it — seriously. Free and good beats $29/yr and good. Close this tab.
Still here? Then you're probably looking at the paid tier, and that's where the two apps actually diverge — not on "local vs cloud," but on price and on what happens to your words after they're transcribed.
| Vokk | Superwhisper | |
|---|---|---|
| per month | $4 | $8.49 |
| per year | $29 | $84.99 |
| own it outright | $49 (first 500) | $249.99 |
| free tier | 14-day trial, no card | permanent free tier |
| on-device by default | yes | yes |
| LLM rewrites your text | never | optional, via modes |
| meeting recording | no | yes |
| platforms | macOS 13+ (Apple Silicon) | macOS, Windows, iOS |
| account required | no | — |
Superwhisper figures are from their own public pricing and feature pages as of August 2026. We don't publish latency, RAM, or CPU comparisons against other apps — we have no verified measurements of theirs, and neither does anyone quoting numbers at you.
Price, if you're paying at all. $29/yr against $84.99/yr. $49 once against $249.99 once. The $49 founding-forever license is the first 500 buyers, up to 3 Macs, every on-device feature for life. Optional cloud transcription (Groq) stays subscription-only, because it costs money every time you use it.
It types what you said. No LLM sits between your mouth and the cursor. That's not a limitation we're apologizing for — it's the product. When I dictate a variable name, a flag, a half-formed thought I'm going to argue with in the next sentence, I want that on screen, not a tidied paraphrase of it. Configurable rewriting is a feature until the day it quietly rewrites something you meant literally.
It does one thing. Hold a key — fn on laptops, ⌃⌥ on desktops, customizable — talk, release. Words appear in whatever has focus. There are no modes to configure, no prompt library to maintain, no per-app settings to get wrong. Vokk types locally into the focused window, so it works in any terminal window regardless of what that terminal is connected to.
It works with the Wi-Fi off. whisper.cpp on your Mac's GPU via Metal, well under a second. The airplane test: turn Wi-Fi off, hold the key, it types anyway. No account, no sign-up, no telemetry.
One developer answers the email. I write Vokk by dictating into Claude Code 12+ hours a day — about 2.05M lines and ~3,750 commits in two months, spoken. Support goes to [email protected] and I'm the one reading it. That's a small-shop trade-off in both directions: faster answers, smaller surface area.
Superwhisper is a bigger app that wants to be your voice-to-anything layer, with an LLM on the far end and a free tier that costs you nothing to try. Vokk is a small app that wants to be a key you hold. Both keep the audio on your Mac by default. Pick by whether you want your words edited on the way in — and, if you're paying, by whether $200 of difference on a lifetime license matters to you.