Download harlo

The latest version for Linux (x86-64), free and open source. All releases

Ubuntu 22.04 or later, Debian 12 or later, Linux Mint, Pop!_OS, elementary OS

harlo_version-1_amd64.deb

sudo apt install ./harlo_*_amd64.deb

Fedora 36 or later, openSUSE Tumbleweed, Nobara

Download .rpm

harlo-version-1.x86_64.rpm

sudo dnf install ./harlo-*.rpm
# openSUSE:
sudo zypper install ./harlo-*.rpm

Arch, Manjaro, EndeavourOS, CachyOS, Garuda

harlo-version-1-x86_64.pkg.tar.zst

sudo pacman -U ./harlo-*.pkg.tar.zst

One file that runs on most distributions, nothing to install

harlo-version-x86_64.AppImage

chmod +x harlo-*.AppImage
./harlo-*.AppImage

After installing

On its first start, harlo's setup guide helps you install the agents you want (Claude Code, OpenCode, pi, Oh My Pi) and sign them in. harlo --setup opens it again.
harlo checks for new versions and updates itself with your system's package manager (Settings → General).
Dictation's speech model (148 MB) downloads the first time you use the microphone.
On the iPhone, get harlo from the App Store, then in harlo on your computer: Settings → Devices → Pair a device, and scan the code.

System	Linux, x86-64 with AVX2 (most PCs since 2013), glibc 2.35 or newer
Included	Its own Qt 6, so it runs where the system's Qt is older
Uses	`git`; optionally `ssh` (other machines), `notify-send` (notifications)
Checksums	Each release lists its files' SHA-256 sums in `SHA256SUMS`

AUR: harlo-bin (the release build), harlo (built from source against Arch's Qt), harlo-git (the latest commit).
From source: Rust 1.85+, Qt 6.9+, cmake and clang, then cargo run -p harlo-app --release. See the README.