Superwhisper, at $8.49 a month with on-device processing and custom modes, scores 18/20 on four criteria taken from Co-Active Coaching (listening, self-management, forward-and-deepen, and a safe environment), because it can be told to keep your rambling verbatim. Wispr Flow ($15 a month, 2,000 free words a week) scores 11/20: its auto-editing is the best for email and the worst for thinking. MosMos, launched September 19, 2026, scores 13/20.
I think better out loud than on a keyboard, and for two years I have been trying to get a dictation app to behave like a coach rather than a typist. The distinction matters. A typist wants clean sentences. A coach wants the hesitation before the sentence, because that is where the real answer usually is.
The three tools worth testing this month. Wispr Flow is the cloud-processed leader: Free covers 2,000 dictated words a week on desktop and Pro is $15 per user a month or $144 a year, on Mac, Windows, iOS and Android, with a Mac Notetaker for meetings and auto-editing that removes filler and reformats text for the app you are in. Superwhisper runs on-device, with a free tier, Pro at $8.49 a month, a yearly plan and a lifetime licence, on Mac, Windows, iOS and Android, and its defining feature is custom modes with your own prompt controlling what happens to the transcript. MosMos is the newcomer, a native macOS voice workspace that hit Product Hunt on September 19, 2026: press Fn anywhere to speak, it adapts the writing style to the app, remembers your vocabulary, records multi-speaker meetings into speaker-separated transcripts with action items, and has an Ask MosMos mode for questions and web searches with reviewable sources. Its pricing was not published on the launch listing when I checked.
For criteria I used the five contexts of the coach from Co-Active Coaching, collapsed to four because intuition is not something I can score in software. Listening: does the tool hear what I meant, including my names and jargon, on the first pass? Self-management: the Co-Active coach "sets aside personal opinions, preferences, pride," and the light "should be shining on the coachee, not the coach"; for a dictation app that means not rewriting my half-formed thought into confident prose it never had. Forward and deepen: can it push the thinking, by returning questions or actions, not just text? And the coaching environment, "safe enough to take risks," which for a founder dictating a board memo or a co-founder conflict means on-device processing.
| Co-Active criterion | Wispr Flow | Superwhisper | MosMos |
|---|---|---|---|
| Listening: accuracy on names and jargon | 5 | 4 | 4 |
| Self-management: leaves the raw thought alone | 2 | 5 | 3 |
| Forward and deepen: returns questions or actions | 2 | 4 | 4 |
| Safe environment: on-device processing | 2 | 5 | 2 |
| Total out of 20 | 11 | 18 | 13 |
The scores need explaining because Wispr Flow is the better product for the job most people buy it for. Its listening row is the best of the three: it learns names and jargon and its cloud models are simply more accurate than the local Whisper variants on my machine, and its context-aware formatting makes a dictated Slack message read like a typed one. That same strength is why it fails the self-management row for thinking. When I say "I think we should, no, actually the problem is the second customer, not pricing," Wispr Flow gives me a tidy sentence about the second customer. The reversal, which was the insight, is gone. Superwhisper wins that row because I built a mode whose prompt is literally "keep everything, including false starts, mark reversals with an arrow," and another mode that returns the transcript plus three questions a coach would ask; that second mode is why it also wins forward-and-deepen, and on-device processing is why it wins the environment row outright. MosMos sits in between on my three days with it: the Fn-anywhere capture is the fastest of the three, the meeting action items and Ask MosMos with sources are a real deepen feature, but it polishes by default like Wispr Flow, and I could not find documentation of where the audio is processed, so it scores as cloud until it says otherwise.
What I actually run: Superwhisper in a "raw" mode for thinking sessions, ten to fifteen minutes of walking and talking that land as a messy transcript in a daily note, then a second pass where I paste it into Claude and ask only for the contradictions. Wispr Flow stays installed for email and messages, where I want a typist. I would move to MosMos for meetings if a pricing page and a processing statement appear, because a single Fn key for both solo thinking and multi-speaker calls is the right shape.
Limits. These are my scores from my use, not a benchmark, and the accuracy gap depends on your accent, microphone and how much jargon you carry. Wispr Flow's Command Mode editing is desktop-only and its Free allowance resets weekly, so heavy talkers hit the wall by Wednesday. Superwhisper's custom modes are a power-user feature; a founder who will never write a prompt gets a good verbatim tool and nothing more. And no dictation app fixes the underlying failure, which is dictating a decision instead of thinking one through: the tool that keeps your reversals only helps if you go back and read them.
Sources: Wispr Flow Help Center, "Flow plans and what's included," plan prices and free allowances (docs.wisprflow.ai); Superwhisper homepage, plan tiers and Pro pricing (superwhisper.com); Product Hunt, "MosMos: Voice writing that works before, during, and after meetings," launch listing, September 19, 2026 (producthunt.com/products/mosmos); Voibe, "Wispr Flow vs Superwhisper: Honest Comparison Guide (2026)," a competitor's comparison used for platform and processing details (getvoibe.com); Henry Kimsey-House, Karen Kimsey-House, Phillip Sandahl and Laura Whitworth, Co-Active Coaching, 4th edition (Nicholas Brealey, 2018).
Related: How to Find Your Passion · Best Self-Improvement Books · How to Make Better Decisions · Why Exploration Is Important for Success