Download

Windows (stable) and macOS (Beta)

Windows: unzip it anywhere and double-click FaceNest.exe — no installer, no admin rights, no account. macOS: a Beta for Apple Silicon, unsigned — read the first-launch steps below.

Windows 10 / 11 · 64-bit

Portable build — unzip and run. No installer, no account, no admin rights.

Download FaceNest v1.0.0

After downloading, verify the SHA-256 checksum (on Windows: certutil -hashfile <file> SHA256).

Third-party licences (FFmpeg / LGPLv3): This software uses code of FFmpeg licensed under the LGPLv3 and its source can be downloaded here. We ship FFmpeg unmodified, built from that exact source archive with LGPL-compatible options; the build script (configure line included) sits next to the source archive. The package also contains dav1d (BSD-2-Clause). The complete third-party notices are inside the download, in the licenses/ folder. FFmpeg is a trademark of Fabrice Bellard, originator of the FFmpeg project; FaceNest is not affiliated with, or endorsed by, the FFmpeg project.

Windows will warn you the first time: This build is not code-signed, so SmartScreen may show a blue "Windows protected your PC" panel. To continue: click More info → Run anyway. The checksum above lets you verify the file yourself first. We never ask you to turn off your antivirus.

Three steps to a running app

  1. Unzip it to a folder you like

    Non-ASCII characters in the path are fine. Make sure the _internal folder is extracted alongside FaceNest.exe.

  2. Double-click FaceNest.exe

    No installation, no admin rights, and nothing is written to the registry.

  3. Face search works immediately

    The face models and the detailed-tagging model ship inside the download, so face search works offline from the very first launch. The two larger models — label search and semantic search — download when you first use them (about 2 GB each). Stay online the first time.

FaceNest showing indexing progress on first run
The first semantic index build takes a while and shows a progress bar — face and tag search keep working throughout.

System requirements

ItemRequirement
Operating systemWindows 10 / 11 (64-bit) — verified on Windows 11 24H2
Operating system (macOS)macOS 14 (Sonoma) or later · Apple Silicon (M1 and newer) · Beta. Intel Macs are not supported — upstream PyTorch no longer publishes Intel-macOS wheels
Disk space — WindowsDownload 6 GB; about 10 GB after unpacking, because the app ships with its core models built in
Disk space — macOSDownload 850 MB; about 1.7 GB installed. Your library and the optional models live in ~/Library/Application Support/FaceNest/
Optional modelsDownloaded on first use: up to roughly 4 GB in total, about 2 GB each
Graphics cardOptional. It runs without a discrete GPU — semantic search is just noticeably slower. With a GPU the backend is picked automatically: CUDA (NVIDIA) → DirectML (other Windows GPUs) → CoreML / MPS (Apple Silicon) → CPU
Memory8 GB or more recommended
NetworkNeeded only to download models, to activate, and for the periodic licence check-in. Everyday browsing and search work fully offline

The first run is slower: The larger optional models download on first use (a few hundred MB up to about 2 GB each), and the first semantic index build takes a while too — there is a progress bar for both. How long depends on your hardware, and we do not promise a number.

macOS (Beta) and Linux

macOS 14 (Sonoma)+ · Apple Silicon

Beta · unsigned and not notarised — read the first-launch steps below before you open it.

Download for macOS (Beta)

After downloading, verify the SHA-256 checksum (on macOS: shasum -a 256 <file>).

macOS (Beta) — on a real Mac

Not a mock-up: these are that build running on an Apple Silicon Mac.

Found a bug? Tell us — that is how this gets better

The fastest route is inside the app: Open Settings → Diagnostic logs, write a short description of what went wrong, and click Upload to support. ⚠️ Put an email address in the description if you want an answer — the upload itself is anonymous, so without one we can only look at the logs. You can also just email support@facenest.net.

Next steps

Something not working?

Check the FAQ first, or just email us.