Get the app

First launch

On first run, Abliteration shows a "Your rig" popup with your RAM, CPU, GPU, and disk, plus what model sizes fit.

Short answer

On first run, Abliteration shows a “Your rig” popup with your RAM, CPU, GPU, and disk, plus what model sizes fit. Click Understand to dismiss; it never shows automatically again. Then add your first model and start chatting. The behaviour is the same on Windows and macOS. The Mac build is signed by Catyonic OÜ and notarized by Apple, so no Gatekeeper warning appears on first open.

Detailed explanation

What happens the first time you open Abliteration:

  1. The app reads your hardware (RAM, CPU, GPU, disk) - takes a second or two on a background thread.

  2. A “Your rig” popup appears with three lines:

    • Line 1 - Your PC: {CPU name}, {N} GB RAM, {GPU phrase} (total RAM installed, GPU informational only)
    • Line 2 - Free right now: {X.X} GB
    • Line 3 - You can run: {tiers based on your RAM} (green/amber/red per tier)

  3. Bottom of the popup: a note saying you can always find this again in Settings → Hardware.

  4. One button: Understand. Click it - popup closes.

The popup is marked as shown in settings.json (HardwareIntroShown = true) so it doesn’t auto-appear again on future launches. You can always open Settings → Hardware to see the same info.

Next steps after the popup:

  • On first launch (no models installed), the Models screen opens beneath the popup - you land there after clicking Understand.
  • The Models tab is empty until you add a GGUF.
  • Once you have a model, the drawer on the left starts collecting chats. Until then it shows just “New chat”.

Why it matters

The first-launch popup is Abliteration’s way of saying “here’s what I see, here’s what will realistically run on your PC.” It sets expectations before you start downloading models - no wasting time on a 30 GB model your 16 GB laptop can’t handle.

In Abliteration specifically

The popup is dismissible only via Understand - no ✕ close button, so you can’t accidentally miss reading the info. Everything is theme-aware (light/dark).

If you install a bigger GPU or upgrade RAM later, re-check via Settings → Hardware - the numbers update whenever you open that screen.

Nothing is uploaded, nothing is sent anywhere. The hardware read runs entirely against local OS APIs on both platforms and never touches the network.

Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Get the app Browse the catalog
Open in Abliteration