Download

Free. Yours to keep.

Runs as an ordinary user. Works offline from the first launch. No account, ever.

Version

AllInsight 1.0.0

Free. Runs as an ordinary user. Works offline from the first launch.

Windows

Windows 10 (1809 or later) and Windows 11, x64

InstallerAllInsight_1.0.0_x64-setup.exe 3.2 MB PortableAllInsight.exe 10.8 MB
Verify the download

The installer is not code-signed yet, so Windows shows “Windows protected your PC”. Choose More info, then Run anyway, after checking the hash:

Get-FileHash .\AllInsight_1.0.0_x64-setup.exe -Algorithm SHA256
Installer
published with the release
Portable
published with the release

Licence MIT  ·  Third-party licences

Windows requirements

Windows 10 or 11, x64.

Windows 10 version 1809 or later, or Windows 11. The installer adds the Microsoft Edge WebView2 runtime if it is missing; Windows 11 already has it. That one step needs a connection. The application itself never does.

Linux requirements

Any modern distribution.

x64 with WebKitGTK 4.1, preinstalled on GNOME and KDE desktops; the .deb and .rpm pull it in. Optional: udisks2 for drive health, and libayatana-appindicator3 for the tray icon.

Optional

Add a local model in four steps.

AllInsight works fully without one. A model only adds conversation, and you do this once, deliberately.

  1. Get the engine

    Download a llama.cpp release and put llama-server in %LOCALAPPDATA%\AllInsight\engine\ on Windows or ~/.local/share/AllInsight/engine/ on Linux.

  2. Get a model

    Any instruction-tuned GGUF. AllInsight AI suggests a size that fits your memory and disk; see the AI page.

  3. Import it

    Open AllInsight AI and choose Import a GGUF model. AllInsight checks the file really is a GGUF before copying it.

  4. Load it

    Press Load. The engine starts as a separate process, so a model that misbehaves cannot take the interface with it.