Skip to content

Mintu v0.1.2: Modular Island, Custom Storage Selection & UI Overhaul

Choose a tag to compare

@HuseynDeveloper HuseynDeveloper released this 18 Oct 10:19
· 9 commits to master since this release

Mintu v0.1.2 — UI Fixes, Modular Island & Storage Selection

Mintu is an open-source, mint-green desktop AI companion in a Dynamic Island-style widget, featuring local offline LLM chat (via llama.cpp), optional cloud AI providers, memory, tools, and interactive status widgets.

What is New in v0.1.2

  • Custom Disk & Storage Selection: Users can now select any disk drive (e.g. D:) or custom directory to download and store local LLM models, solving low-space constraints on system drives.
  • Onboarding and Scroll Fixes: Resolved button clipping on the onboarding screen; added smooth scrolling and adaptive sizing for model confirmation actions.
  • Redesigned Message Actions: Replaced cramped edit controls with a clean action row (Edit, Copy, Regenerate, Remember) and inline dark-glass editor (Ctrl+Enter to save, Esc to cancel).
  • Provider & Model Badge: Clear visual distinction in the chat header between Local models (Mint green) and Cloud providers (Amber badge with cloud icon), accompanied by a fast model switcher.
  • Unified Notification System: Fixed duplicate icons and stacking bugs; Mintu now dynamically wakes up and reacts when notifications arrive.
  • Island Modules: Built-in Focus Timer with circular progress ring, Stopwatch, and Live System Status widgets.
  • UI Stress Mode: Diagnostic testing tool ensuring zero letter-wrapping and no layout clipping across languages and DPI scaling.

Installation and Usage

  1. Download Mintu-v0.1.2-Windows.zip from the Assets section below.
  2. Extract the archive to any folder on your computer.
  3. Launch mintu.exe.
  4. The Mintu island widget will appear at the top center of your screen. Click it to expand or chat.

System Requirements

  • Windows 10 / 11 (64-bit)
  • Microsoft Edge WebView2 Runtime (installed by default on modern Windows)
  • The local inference engine (binaries/llama/) is bundled directly in the release zip.