Question · Dictation software
What is the Hex dictation app?
HEX is an open-source dictation app for Apple silicon Macs running macOS 15 or newer. You hold a shortcut, speak and release it, and it transcribes on your Mac. The older Swift app called Hex was archived on 14 October 2026, so it is a different project.
Updated October 16, 20264 min read8 cited facts, each traced to its source
HEX is a dictation app for the Mac. Its README describes it as local-first voice dictation for Apple silicon Macs running macOS 15 or newer 1. Its code is published under the MIT License 2. Because an older app with a similar name exists, check which one you have before you install anything. This page covers the current HEX, what it needs, what its optional cloud step does, and whether it fits your Mac. The open-source dictation guide compares HEX with the other open-source projects.
You hold a shortcut, speak and release, and HEX transcribes on the Mac
The README's usage is short. You hold a shortcut, speak, and release it. HEX then transcribes on your Mac and pastes the result into the app you are using 3. The speech model is a file you download once. During setup you download a local speech model, and the README says ordinary dictation does not need OpenCode or a cloud transcription service 4. That is the main difference from a cloud app: the words are converted to text on your machine.
The old Swift Hex app is archived, so the current HEX lives at a different address
Two projects have used the name Hex. The original Swift app was archived by its owner on 14 October 2026 and is now read-only 5. An archived repository does not receive fixes, so anyone still running the old app will not get new updates from it. The current project is HEX at github.com/anomalyco/hex, and the facts on this page come from its README and license. If the app you installed came from the old repository, uninstall it and get HEX from the current project instead.
Models are chosen per language, and the cloud step is optional
The app's model menu lets you switch between the speech models you have downloaded. The README's example is English with Parakeet and French with Whisper 6. The model choice is yours, and the README gives that pairing as an example rather than a default.
The optional text processing is a separate decision. When you turn on post-processing with OpenCode in Modes, the README says HEX sends the transcript, the foreground application and the Brave website hostname to the model provider you configured 7. The README also says ordinary dictation does not need this feature. Leave it off if you do not want those details to leave your Mac.
The Linux version is a beta with fewer features
HEX also has a Linux beta. The README says the beta has a smaller feature set and requires a source build or Nix 8. If you are not comfortable building software from source, the Linux beta is not a good first choice.
HEX fits Apple silicon Macs on macOS 15 or newer, and not other machines
The fit depends on your hardware and your software, and the README states both limits.
- Apple silicon Mac on macOS 15 or newer: this is the setup HEX is built for. Start here.
- Intel Mac: the README names Apple silicon Macs, so an Intel Mac is outside the stated scope.
- macOS 14 or older: the README requires macOS 15 or newer, so update first or choose another app.
- Linux: the beta is the only route. It suits people who can build from source.
- Cloud rewriting: you can turn it on, but then the text and app details go to the provider you chose.
If you need a company behind your support and a single installer, a paid app may suit you better. The dictation software overview compares the paid options by platform, price and processing.
Steps to confirm your Mac and install HEX
- Open the Apple menu and choose About This Mac. Check the chip line. An Apple silicon Mac shows an Apple M chip, not an Intel processor.
- In the same window, read the macOS version. It must be macOS 15 or newer.
- Download HEX from the Download for macOS link in the project's README at github.com/anomalyco/hex.
- During setup, download one speech model for the language you dictate in.
- Leave text processing in Modes off unless you choose it, then test the app on a paragraph you have written yourself.
For a broader method of testing any dictation app, see how to test dictation software. If you are comparing Handy, the other open-source Mac-and-Windows project, read what is the Handy dictation app. For the license and price of VoiceInk, another open-source Mac app, read is VoiceInk free and open source. For Mac dictation in general, see our offline dictation on Mac page, and the dictation software entry defines the term.
Evidence
Every number on this page, with where it comes from. Open a fact to see the exact words in its source.
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
Go deeper
Related questions
- Can you run voice recognition software on a Raspberry Pi?
- Is VoiceInk free and open source?
- What is the Handy dictation app?
- Can you upgrade an old Dragon NaturallySpeaking to version 15 or 16?
- Does Apple have a voice to text app?
- Does Dragon NaturallySpeaking need an internet connection?
- Does Dragon NaturallySpeaking use AI?
- Does Dragon NaturallySpeaking work over Remote Desktop?