Fixes
- Fixed an incorrect phi3.5 model tag causing local model downloads to fail (#690, #691) — thanks @sinnrrr
- Fixed cloud transcription credentials not being passed through for audio imports (#689, #688) — thanks @gthay
- Updated AssemblyAI integration to use their latest default speech model instead of a pinned older one (#692) — thanks @gthay
Housekeeping
- Homebrew cask updated to 1.84.2 (#693)