Updated September 10, 2026 · By Sumbat.T
Handy vs Wispr Flow: Free Offline or Paid Cloud in 2026?
The short answer
- Handy transcribes. Wispr Flow rewrites. Handy gives you back what you actually said, processed on your own machine. Wispr Flow sends the audio to the cloud, strips the filler words and reshapes the sentences. Everything else follows from that one difference.
- Handy is free and it is genuinely free. MIT licensed, no tiers, no account, no cap. Wispr Flow Pro is $15 a month, or $12 billed annually, which is $144 a year. Read from the vendor's own pricing page on 10 September 2026.
- Two things the rest of page one gets wrong. Handy is at 31,321 GitHub stars, not the ~19,900 the top results still quote, and it ships a post-processing toggle rather than requiring you to wire up an external LLM API.
- Only one of them works offline. Handy runs Whisper or Parakeet models locally, so nothing leaves your computer. Wispr Flow cannot type a word without a connection.
- If what you want is to decide how your speech gets rewritten, neither is built for it: one does nothing, the other does a fixed thing. BlabbyAI custom modes are rules you write in plain English, on Windows and in Chrome, at $8.49 a month.
The third name in this test
You write the rule, not the vendor
Handy hands back raw text and Wispr Flow applies one fixed cleanup. BlabbyAI custom modes are plain-English instructions you author yourself, on Windows and in Chrome, for $8.49 a month with a free tier that needs no card.
Best for, in one line each
Every price and specification on this page was read first-hand from the vendor's own pricing page, the project's own README or the GitHub API on 10 September 2026, not copied from another comparison.
Best for keeping audio entirely on your own machine: Handy
Handy runs Whisper or Parakeet models locally and sends nothing to a server, so it is the only tool in this comparison that keeps dictation working with the network off.
Best for turning rambling speech into finished prose: Wispr Flow
Wispr Flow runs an editing pass in the cloud that removes filler words and restructures sentences, which is work Handy leaves to you.
Best free option with no word cap at all: Handy
Handy is MIT licensed with no paid tier and no usage limit, while the Wispr Flow free plan stops at 2,000 words a week on desktop.
Best for writing your own rules for how speech is rewritten: BlabbyAI
BlabbyAI custom modes are plain-English instructions you author yourself, so the cleanup follows your rule rather than a fixed setting someone else chose.
Best for dictating on an Android phone for free: Wispr Flow
The Wispr Flow free tier is unlimited on Android while capping desktop at 2,000 words a week, and Handy has no mobile build at all.
Best for a Windows PC where cost is the deciding factor: BlabbyAI
BlabbyAI is $8.49 a month against the $15 Wispr Flow charges, runs natively on Windows, and needs no per-machine model download or GPU.
Side by side
| Tool | Price | Where it runs | What you get back | Platforms |
|---|---|---|---|---|
 Handy | Free, MIT licensed | Fully local, works offline | Faithful transcript, post-processing is a toggle | macOS (Intel and Apple Silicon), x64 Windows, x64 Linux |
 Wispr Flow | $15/mo, $12/mo billed annually | Cloud, needs a live connection | Rewritten prose, filler words removed | Mac, Windows, iOS, Android |
 BlabbyAI | $8.49/mo, no annual lock-in | Cloud, History keeps audio on your own disk | Custom modes: you write the rewrite rule | Windows, Chrome extension, Linux |
The row that decides most of this is the third one. A tool that runs on your own machine cannot afford a large editing model, because the model has to fit on your laptop and finish before you lose patience. A tool that runs in the cloud can afford one, and charges you for it. Handy and Wispr Flow are not really competing on quality; they sit on opposite sides of that trade, and the right question is which side of it your work falls on.
Handy: a microphone that types, and nothing else
Handy describes its own ambition more clearly than any review of it does. From the project README: "One tool, one job. Transcribe what you say and put it into a text box." And a line below that, the sentence that explains every design decision in the app: "Handy isn't trying to be the best speech-to-text app, it is trying to be the most forkable one."
Mechanically it works the way most desktop dictation tools do. You hold a configurable shortcut, or tap it to toggle, speak, and release. Silence is trimmed using Silero voice activity detection, then the audio goes to a model you choose: a Whisper model (Small, Medium, Turbo or Large) with GPU acceleration where available, or Parakeet V3, a CPU-optimised model with automatic language detection. The text is pasted into whatever app you were already in. The whole path runs on your machine.
31,321 stars
Read from the GitHub API on 10 September 2026, with 2,835 forks. The repository was created on 13 February 2025 and was last pushed to on the same day this article was checked.
MIT licensed
One of the most permissive open-source licences. Commercial use, modification and redistribution are all allowed without payment or permission.
x64 only on Windows
The README lists macOS on Intel and Apple Silicon, x64 Windows and x64 Linux. There is no ARM Windows build, so Snapdragon laptops are not covered by the official releases.
157 open issues
Normal for a project of this size, and a fair picture of what community-maintained support looks like when something breaks.
Handy is a Tauri app, and that is why it is small
Two things almost every page about this comparison gets wrong
Both of these are checkable in about a minute, and both are currently being repeated by the highest-ranking results on this query, including Google's own AI Overview. They matter because each one, if you believe it, points you at the wrong tool.
Widely repeated
Handy has around 19,900 GitHub stars.
What the source says
It has 31,321, with 2,835 forks. The project has added more than eleven thousand stars since the pages quoting the older figure were written, and it was last updated on 10 September 2026.
Source: GitHub API, repository cjpais/Handy, read 10 September 2026.
Widely repeated
Handy cannot remove filler words or auto-edit unless you wire up an external LLM API yourself.
What the source says
Handy ships a built-in post-processing mode. It is a separate bindable action from plain transcription and is exposed on the command line as handy --toggle-post-process, documented alongside --toggle-transcription and --cancel.
Source: Handy README, cjpais/Handy on GitHub, read 10 September 2026.
The second correction does not turn Handy into Wispr Flow. Handy's post-processing is a toggle you switch on, and Wispr Flow's editing is the product. But the difference between "this app cannot do that" and "this app does that when you turn it on" is large enough to change which one you download, and right now the search results are telling you the first one.
Wispr Flow: an editor that listens, for $144 a year
Wispr Flow does the thing Handy deliberately does not: it takes the rough transcript and makes it read like something you wrote on purpose. Filler words go, half-finished sentences get closed, and the output arrives already formatted for whatever you appear to be writing in. That pass happens on a server, which is both the reason it works and the reason the app is useless without a connection.
Its scope is also wider than dictation. The paid product bundles a meeting Notetaker that identifies speakers, connects to your calendar and Slack, and exposes notes to Claude and ChatGPT over MCP. One caveat worth knowing before you pay: on Wispr Flow's own pricing page, the Notetaker is marked "Mac only for now", so a Windows subscriber pays the same $15 for a product missing that feature.
| Plan | Price | What that buys | Notes |
|---|---|---|---|
| Handy | Free | No tiers, no account | MIT licensed. 31,321 GitHub stars, 2,835 forks, 157 open issues as of 10 September 2026. |
| Wispr Flow Free | $0 | 2,000 words/week desktop | 1,000 words a week on iPhone, unlimited on Android. Notetaker is Mac only. |
| Wispr Flow Pro | $15 / mo | $12/mo billed annually ($144/yr) | Unlimited dictations, advanced models, longer meeting retention. |
| Wispr Flow Growth | from $18 / mo | per user | Team management and admin controls. Omitted from most comparisons of this pair. |
| BlabbyAI Free | $0 | 60 credits weekly | No card required. Windows app and Chrome extension both included. |
| BlabbyAI | $8.49 / mo | No annual commitment | Unlimited dictation, custom modes, History keeps audio on your own disk. |
The free tier is three limits, not one
There is also a tier above Pro that most write-ups of this pairing leave out entirely, the AI Overview on this query included: Growth and Enterprise, starting at $18 per user per month. If you are pricing this for a team rather than for yourself, that is the number that will apply to you.
The question underneath this comparison
Most comparisons of these two frame the choice as free against paid, or private against convenient. Those are real axes, but they are not the one that will annoy you six weeks in. The one that will is this: who decides what your speech turns into?
With Handy, you do, in the sense that nobody does. You get the transcript and any tidying is yours to do afterwards, or to configure through the post-processing toggle. With Wispr Flow, the vendor does. Their cleanup pass has a house style, and it is a good one, but it is fixed. When it renders your bullet points as a paragraph, or formalises a message you wanted to sound casual, there is no lever to pull.
A third answer: write the rewrite rule yourself
This is the gap BlabbyAI is built around. A custom mode is a free-form instruction, written in plain English, that gets applied to whatever you just said. Not a setting you toggle and not a preset you pick from a list: a sentence you write. "Turn this into short bullet points." "Keep my wording but fix the punctuation and nothing else." "Format this as a commit message." The mode does that, every time, until you change it.
Source: BlabbyAI custom mode editor. The instruction is written in plain English and applied to whatever you dictate next.
The practical difference shows up on the small stuff. If you dictate Slack messages and Jira tickets and patient notes in the same afternoon, all three want different treatment, and a single fixed cleanup style is wrong for at least two of them. Modes let you bind a different rule to each. It works in any app with a text field, because it types into the field rather than integrating with the app, so Teams, Notion, Word, an IDE and an EHR all behave the same way.
On the two axes this comparison actually turns on, it lands in a different place from either. On price it is $8.49 a month against the $15 Wispr Flow charges, with no annual commitment and a free tier of 60 credits a week that needs no card. On where things live, History keeps your audio on your own disk rather than in an account you have to trust. And unlike Handy there is a Chrome extension that needs no download and no admin rights at all, which matters if the machine you type on all day is a locked-down work laptop where you cannot install a desktop app in the first place.
What Handy is genuinely better at
How to decide in twenty minutes
Because Handy costs nothing and both of the others have free tiers, there is no reason to decide this from an article, including this one. What follows is the shortest test that actually separates them.
- Pick fifty sentences you would really dictate. Not a paragraph from a book. Two Slack replies, an email, and something long and rambling that you have not planned out. The rambling one is the test; the tidy one will look fine everywhere.
- Run them through each tool without editing. Paste the raw output somewhere and read all three side by side. Do not fix anything as you go, or you will not be able to tell which tool did the work.
- Count the corrections, not the errors. The number that matters is how many times you had to reach for the keyboard afterwards, because that is the cost you will pay every day.
- Then ask whether the output was wrong or just not yours. If it was wrong, you have an accuracy problem and a better microphone may fix more of it than a different app. If it was accurate but not in your voice, you have a rules problem, and that is the one custom modes solve.
Running both is a legitimate answer
Frequently asked questions
What is the difference between Handy and Wispr Flow?
Handy transcribes what you said. Wispr Flow rewrites it. That is the whole comparison in two sentences, and almost every other difference follows from it. Handy is a free, MIT-licensed desktop app that runs Whisper or Parakeet models entirely on your own computer, so nothing leaves the machine and it works with the wifi off. What comes back is a faithful transcript, including the false starts. Wispr Flow sends your audio to the cloud, runs an editing pass over it, and returns finished prose with the filler words removed and the sentences reordered, for $15 a month. So the question is not really which app is better. It is whether you want a microphone that types or an editor that listens.
Is Handy really completely free?
Yes, and unusually so. Handy is MIT licensed, which is one of the most permissive open-source licences there is: you can use it commercially, modify it, and ship your own fork without paying or asking. There is no paid tier, no account, no usage cap and no telemetry-driven upsell, because there is no company to sell you anything. The costs are real but they are not money. You supply the compute, so a slower machine means a slower transcription; you supply the setup time; and you supply your own support, which in practice means the GitHub issues page. At the time of writing the repository has 157 open issues, which is normal for a project this size and also exactly what community-maintained means in practice.
How many GitHub stars does Handy actually have?
It has 31,321, read directly from the GitHub API on 10 September 2026. This is worth stating because nearly every page currently ranking for this comparison is working from a much older number: the top Reddit thread on the query says 19,900-star, and Google’s own AI Overview repeats roughly that figure. The project has added more than eleven thousand stars since those pages were written and was last updated the same day this article was checked. If you are using popularity as a proxy for whether the project will still be maintained next year, use the current number rather than the cached one.
Does Handy work on Windows?
Yes, but read the fine print on the architecture. Handy’s own README lists its platform support as macOS on both Intel and Apple Silicon, x64 Windows, and x64 Linux. The word to notice is x64. There is no ARM Windows build listed, so if you are on a Snapdragon-based Windows laptop, which is an increasingly common class of machine, the official builds do not cover you. On a normal x64 Windows PC it installs cleanly, and there is a winget package (winget install cjpais.Handy), though the README notes that the winget and Homebrew packages are not maintained by the Handy developers themselves.
Can Handy clean up my speech, or does it only produce raw text?
It can, and this is the single most commonly repeated error about Handy. Google’s AI Overview on this comparison currently states that Handy does not automatically remove filler words or auto-edit unless you set up custom post-processing via an external LLM API. Handy’s own README documents a built-in post-processing toggle, exposed as a command-line flag (handy --toggle-post-process) and as a bindable action, distinct from plain transcription. So the honest version is that Handy ships post-processing as a first-class feature you turn on, rather than something you have to bolt on from scratch. It is still a toggle rather than a set of instructions you write, which is a different limitation from the one usually reported.
What does Wispr Flow cost, exactly?
Pro is $15 per user per month billed monthly, or $12 per user per month billed annually, which works out at $144 a year. Both figures were read from wisprflow.ai/pricing on 10 September 2026. There is a free tier at $0 and, above Pro, a Growth and Enterprise tier starting at $18 per user per month that most comparisons of this pairing omit entirely, including the AI Overview currently sitting at the top of this query. If you are budgeting for a team rather than yourself, $18 is the number that will actually apply to you, not $15.
What are the limits on the Wispr Flow free tier?
Three limits, not one, and they differ by device. Read from the vendor’s own pricing page on 10 September 2026, the free plan gives 2,000 words a week on desktop, 1,000 words a week on iPhone, and unlimited on Android. That last one is genuinely unusual and rarely mentioned: if you dictate mostly on an Android phone, the Wispr Flow free tier is effectively uncapped for you. On a laptop, 2,000 words a week is roughly one long document, so most people who dictate daily will hit it inside two or three days.
Which one is more private?
Handy, without any real argument, and this is its strongest claim. Transcription happens on your own computer using local Whisper or Parakeet models, so the audio never travels anywhere and the app keeps working on a plane. Wispr Flow processes audio in the cloud, which is how it does the editing pass, and it holds SOC 2 Type II and ISO 27001. Those are meaningful controls over how a vendor handles your data, but they are not the same claim as not sending the data at all. If your requirement is that the audio must not leave this machine, that is a requirement only local processing satisfies, and among these two only Handy does it.
Which should I pick if I dictate all day for work?
Start by separating two questions that this comparison usually welds together: where the processing happens, and how much editing you want. Handy answers the first well and the second only with a toggle. Wispr Flow answers the second well and the first not at all, at $144 a year. If what you actually want is control over how your speech is rewritten, rather than a fixed cleanup someone else designed, neither is built for that. That is the specific gap custom modes fill in BlabbyAI: a mode is an instruction you write in plain English, applied to whatever you just said, so you decide whether the output is a tidy paragraph, a bulleted list or a formatted note, at $8.49 a month.
Can I run both?
Yes, and for a lot of people that is the right answer rather than a fence-sitting one. They cost nothing together beyond the one subscription, because Handy is free. A common split is to keep Handy bound to one shortcut for anything sensitive or offline, such as notes about clients, personal writing, or work on a train with no signal, and keep a cloud tool bound to another for the long-form drafting where the editing pass earns its money. The practical limit is that both apps want a global keyboard shortcut, so pick two that do not collide before you install the second one.
Run the same fifty sentences through a third option
60 credits a week, no card, no account to set up. Handy gives you the raw transcript and Wispr Flow gives you its house style. Find out in twenty minutes what your own rule sounds like.