- New "Chapters" AI mode - Fixed an issue with some MP3 files with the Parakeet model

• Parakeet is now about 2x faster • Ready for the iPhone Duo • Benchmark mode is now free for everyone

- Shortcuts and Siri: transcribe a file, grab the text, export subtitles, or open a transcript, all from the Shortcuts app. - Spotlight: search your transcripts by name or by what was said. - Apple Intelligence: when your monthly AI requests run out, Analysis and Clean Up keep working using Private Cloud Compute. - iPad and wide screens now get the two-column layout. - On macOS 27 and iOS 27, drag transcripts to reorder the sidebar.

NVIDIA Parakeet TDT V2 has been added. It's a bit faster than Parakeet V3, but is English-only. Whisper now uses the large-v3 Turbo model, a bit faster than the old Whisper model and requires a new download.

Quick Subtitles 3.0 is a full redesign. The app transcribes audio, so it now looks like an audio tool. Every screen has been rebuilt around a studio hardware look, with depth coming from bevels and recessed wells instead of soft shadows, and there are now two properly designed appearances: a machined dark mode and a crisp white light mode. Gold now means something specific: it marks whatever is live, selected, or in progress. Transcription itself is now a console. The waveform is the progress bar, filling with gold as the file is worked through, with elapsed time and time remaining underneath. When a transcription finishes, the console slides away and the finished transcript rises into place. Also in this release: Put your transcripts in whatever order you like. Drag a card up or down the sidebar and it stays where you put it, and your arrangement syncs across your devices. The Whisper model now loads in seconds, every time. Previously macOS could quietly throw away a per-machine optimization cache and force a rebuild that took a minute or more. Dropping files is more forgiving on the Mac: release an audio or video file anywhere over the window and it imports. Setting up on the Mac now happens in a proper setup assistant instead of a panel floating over an empty app. Three new animations (a resonator, a signal path, and a monolith) join the lineup shown while AI cleanup and analysis work. Plus smaller fixes throughout, including the drag preview when exporting subtitle files and the iCloud sync indicator on iPhone.

Updated UI for sharing transcript files, as well as an improvement to data storage to reduce overall app size on your device.

- Fewer files will get converted before transcription begins - The app will use less space on your device - Generally improved performance - Settings screen clean up (no lost features, just more intuitive layout) - UI refinements

Gemini 3.5 Flash model added as the default for Quick Subs AI subscribers, as well as an option for those bringing their own Gemini API key.

Minor stability improvements.

Analysis feature tries to find actual links for you discussed. Gemini and GPT models all updated to the latest versions. Improved loading indicators and clean up UI.

UI improvements throughout, including better clean up and indicators when third party models are being loaded.

- Improved the clean up UI to be a bit more intuitive. - Added model picker to main view - Improved reliability converting video files

Improved loading the Parakeet and Whisper models to prevent cases where the transcription would not start.

Added Whisper from OpenAI as a new model option. This model processes slower, but is the most accurate local model available. Also made an improvement to the sidebar to look a little nicer.

Added VTT subtitle export support.

Quick Subtitles 2.0 is a major update with a redesigned experience and powerful new features. - Multi-Document Support - All your transcriptions are now saved and organized so you can go back to them anytime - Editable Transcripts - Tap to edit any part of your transcription directly - Clean Up - Find and fix transcription errors with suggested corrections you can review inline (or use YOLO mode to auto-apply them all) - NVIDIA Parakeet Engine - A new speech engine option that's up to 2x faster on Apple Silicon - New App Icon Plus numerous bug fixes and performance improvements.

- Improved UI consistency in live activity - Added an option to sign up for a subscription to use the AI features (bring your own key is still here, nothing is exclusive to the subscription). This aims to give a solution to people who don't want to manage their own API keys - Improved benchmark history design - Improved batch processing design

Live activity and background transcription support is here! Start a transcript and go do other things while it works in the background. - Improved line break logic for SRT exports - Added a benchmark download to the benchmark history page.

Benchmark mode has a history section with basic charts and downloadable data. The UI has been refreshed a bit for the new year.

Added a benchmark mode for premium users (find it in settings). GPT-5.2 added for the analysis feature.

- Added support for Gemini 3 Flash - AI Analysis features render responses as rich text, not raw Markdown

Feature update: Set a default save directory so all subtitle files save to the same place every time Thank you for using Quick Subtitles!

Several OpenAI API updates: - Model picker, supporting the latest GPT-5 models (5.1, mini, and nano) - Ability to verify your OpenAI API key Gemini has better response text when something goes wrong.

1. Added the option to use your own OpenAI API key (this API is much slower than Gemini, so be patient if you use it!). 2. Added option to customize the the analysis prompt to whatever you want.

Working on fixing the issue preventing loading the in-app purchase.