The chat drawer - R, C, and Rᶜ markers

The drawer on the left lists all your chats, grouped by date.

Short answer

The drawer on the left lists all your chats, grouped by date. Chats show a small marker to the left of the title indicating what happened in that chat: R (used a reasoning model), C (used Council mode), or Rᶜ (both - the c sits as a superscript on the R). Titles all line up on the same left edge whether they have a marker or not.

Detailed explanation

Layout. The drawer runs down the left side of the app. Chats are grouped into sections: Pinned (top, if any), Today, Yesterday, and older date buckets. Within each section, chats sort by most-recent-activity first.

Row markers - one glyph per row.

Each chat that has a special property gets a single compact letter marker to the left of the title:

  • R - this chat has used a reasoning model (DeepSeek R1, Qwen QwQ, Qwen3-thinking, etc.).
  • C - this chat has used Council mode at any point.
  • Rᶜ - both. The R is at normal size; the c is a superscript on it, the way you’d write x². Reads as one label, not two.

Rows without any of those properties carry no marker - just the title.

Pinned chats have no separate icon. They already live under their own PINNED section at the top of the drawer, so a per-row pin glyph would just repeat what the section header already says. Pin/unpin is done from the row’s ⋯ menu.

Alignment rule. Marker or no marker, every chat title starts at the same X position. The marker occupies a small fixed-width column to the left of the title; when it’s absent, that column stays empty and the title still starts where you expect. Scanning the drawer, all titles form a clean vertical line.

Clicking a chat. Opens it in the main view. The current chat is highlighted in the drawer.

New Chat button. At the top of the drawer. Creates a fresh empty chat with the currently-selected model in the model selector.

Why it matters

Once you have 20+ chats, being able to spot at a glance “which of these has reasoning content” or “which are Council debates” saves opening chats to check. Grouping pinned chats under their own section (instead of tagging every row with a pin icon) keeps the drawer legible - the section header carries the signal without the per-row noise.

In Abliteration specifically

The marker is rendered in the drawer’s on-surface color (dark on the light-themed drawer, light on the dark-themed drawer) - not in an accent color, on purpose. It should read as a subtle label, not a badge.

The Rᶜ composite uses a superscript layout on the c (about 60% of the R’s size, baseline-raised). This keeps the ambiguity of “RC” - two independent letters - away. If Inter’s built-in superscript variant renders on your Windows install it’s used verbatim; otherwise the app draws a manually-offset second glyph to the same effect.

The ⋯ button in the top-right of each row opens the Pin/Rename/Delete menu. There is no right-click handler on rows - only the ⋯ button triggers the menu. Pinning a chat moves it into the PINNED section immediately; unpinning moves it back to whichever date bucket it belongs to.

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.

Open in Abliteration