Highlights

  • Native macOS app with model library sidebar and one-click pull.
  • Faster Metal inference path for Qwen and Llama 3 variants on Apple Silicon.
  • Drag-and-drop document chat without leaving the desktop shell.

View release on GitHub →