Superwhisper launches on Linux, and DHH confirms it ships as Omarchy 4.5’s default dictation
The voice-to-text app that Omarchy 4.5 already picked as its dictation default is now publicly available on Linux, and the decision to start on cloud processing has people arguing in DHH’s replies.
Image: Superwhisper (superwhisper.com)
Superwhisper is officially on Linux. “Introducing Superwhisper for Linux 🐧 Available now,” the company posted on X on November 3, 2026. Hours later, DHH quote-posted the announcement with the news Omarchy users care about: “Superwhisper is shipping as the default dictation setup in Omarchy 4.5.”
The Superwhisper for Linux page lists Omarchy 4.0 or later among supported systems, next to Arch, CachyOS, Ubuntu 24.04+, Linux Mint 22+, Debian 13+, Fedora 40+ and openSUSE Tumbleweed, and adds that Hyprland works “on any distro, Omarchy included.” It “needs Wayland (Hyprland, KDE Plasma or GNOME) and systemd,” comes in x86-64 and ARM builds, and installs with curl -fsSL https://tux.superwhisper.com/install.sh | sh. The page shows Cloud, Offline and Offline Lite modes, so local processing is there too.
DHH’s pitch is all about getting started fast: “The free 10k word cloud option means you can get started without even having to download a model. Then you can switch to local-model dictation later if you like. It’s so good!” The Linux page also has an FAQ entry on what happens after 10,000 Cloud words, a question several people asked under DHH’s post too. Superwhisper’s own reply to DHH: “We can’t wait for 4.5 🙂”
The default itself isn’t new. DHH merged “Ship Superwhisper cloud dictation by default” into the 4.5 branch on October 8, 2026, adding superwhisper-bin to the ISO’s base packages and making superwhisper the value in ~/.config/omarchy/defaults/dictation. Because Superwhisper already defaults to cloud processing, the pull request says no model download or first-run setup prompt is needed, and existing users’ backend choices stay as they are. Voxtype and, coming in the next version, Hex remain the alternatives.
The last few days went into making that default actually work out of the box. According to a pull request merged on November 1, nothing ran Superwhisper’s per-user setup on a fresh install, so “the service, shortcuts and bar panel” only got configured if you picked it again from Setup > Defaults > Dictation. First run now does it unattended in the background. The same evening, an ISO fix stopped aarch64 images from leaving the package out: “An N1x install from today’s image came up with every default package except Superwhisper.”
For machines that already came up half-configured, a migration merged on November 3 finishes the job. If Superwhisper is your selected backend and its setup never ran, it installs the package if it’s missing, runs the setup and fixes the shortcuts, with no settings window or notification. Installs using another backend are skipped. That fix is coming in the next version.
Not everyone loves a cloud default. “Cloud shouldn’t be the default for systems that support fast local inferencing,” @squaredcurve replied. @j3g argued that “local first is the strategy for this project,” and @danexpo said “a cloud solution as a default just doesn’t feel right,” preferring a local model, “even baked into the image.” Adam Zmenak wrote that he usually loves everything Omarchy does but isn’t on board “esp when there are already good local model options that run on very modest hardware.” @slashmili welcomed the Linux version as “more options,” while questioning a cloud default “for such an intimate app that you talk all day.”
Others think DHH has the order right. “Free cloud tier first and local models later is a smart default, because most people won’t try dictation if step one is downloading a model,” Jon Cheney replied. Either way, the cloud is only where you start: as DHH put it, you can move to local-model dictation whenever you like.
Source: https://x.com/superwhisper/status/2117728892782010435
