V0.5.3-BETA

Download & build

Installers (Windows, Linux AppImage, Arch package) are unsigned for now, there's no auto-updater, no download counter to impress you with. It's beta — here's how to get it running today, and what to expect if you do.

LATEST RELEASE

V0.6.1-Beta

Pre-built binaries, straight from the GitHub release page.

All releases →

OR BUILD FROM SOURCE

Linux

Primary dev platform. Clone and run from source for the very latest.

          git clone https://github.com/Sight8218/OPNduck.git
cd OPNduck
# see README.md for current build steps
        

Windows

Supported, less battle-tested than Linux right now. Same source build.

          git clone https://github.com/Sight8218/OPNduck.git
cd OPNduck
# see README.md for current build steps
        

Build steps are changing fast during beta and will go stale here faster than in the repo — the README is the source of truth, not this page.

SYSTEM REQUIREMENTS

GPU

NVIDIA, AMD, or Intel — used when present. Not required.

No GPU

Falls back to CPU. Slower — physics still applies, subscriptions don't change it either.

RAM

8GB minimum, more headroom helps with larger source files.

Disk

Room for AI models plus your working files. Exact footprint is still moving.

OS

Linux (primary), Windows (supported). No macOS build.

RELEASE NOTES

v0.6.1-beta

A quick follow-up to v0.6.0. The assistant can use the internet (off by default) and change settings and the theme when you ask, and shows what it is doing. Fixes the local model running out of memory in Chat. Tidier Settings and a more compact Home. Still a Beta: expect bugs and please report them.

v0.6.0-beta

The biggest update yet, and possibly ever, and it might be super buggy. Recipes, Send to OPNduck, AI chat and tools, Colour Grade, Background Remover, Dubbing, OPNduck Server, an Android app and a macOS beta. Expect bugs and please report them.

v0.5.3-beta

Optional local MCP server so AI apps (Claude Code, Codex, Gemini CLI, Cursor and more) can start and watch jobs. Animated menu icons. Something hiding for the curious.

v0.5.0-beta

The performance update: PyTorch streaming + NVENC video engine, chunked long videos, trimmer with a visual editor, queue controls, watch folder, notifications and tray, live logs, 24 Omarchy palettes, Electron 45.

v0.4.0-beta

The first beta with all the tools working: downloader, converter, upscalers, interpolation, subtitles and AI audio.

v0.2.1-alpha

Patch: the packaged app was effectively unusable on first install — navigation didn't work and social icons were missing, because React Router's BrowserRouter can't resolve routes when the app loads from disk instead of a real server. Fixed by switching to HashRouter and correcting a few hardcoded asset paths.

v0.2.0-alpha

First release with a real, working feature: the Downloader now drives an actual yt-dlp process — video/audio quality selection, size estimates, thumbnails, and a full download queue with cancel. Other tools are still stubs.

v0.2.0-pre-alpha

Rebuilt the whole OPNduck shell, revamped the navigation and UI/UX, and wired up the Glass / Monochrome theme system and settings. Shell-only — the tools weren't functional yet.

v0.1.0-pre-alpha

First public commit. Core pipeline scaffolding — not yet feature-complete, not yet stable. A "come watch it get built" release, not a "come use it" one.

Everything else happens in the commit history — follow the repo for the real feed.

      ← back to OPNduck
    

      GitHub
      GPL-3.0 License
    