Manuwave Studio 3.26.0

Added

• Choose the language model. ⚙️ → Language model for brief and lyrics lists the models of the app's own
  Ollama and — if you already run Ollama yourself — yours as well. Your own models are used directly, nothing is
  downloaded a second time. Automatic stays the default. The language model writes style tags, lyrics and
  structure; the music itself always comes from ACE-Step 1.5.
• Thinking models answer straight away. Models with a thinking phase are asked without it — a brief takes
  seconds instead of minutes.
• If the chosen model is not reachable (your Ollama is not running), the best other one writes and the setting
  says so; your choice comes back by itself once Ollama runs again.

Fixed

• The library folder, the pairing command, the image-model hint, the model-server error and the proof-of-origin
  PDF named a fixed folder of an older layout. They now show the folder of your install (~/ManuwaveStudio,
  %USERPROFILE%\ManuwaveStudio on Windows).