Repository navigation
Releases
Release v5.0.2
- Improved video format selection: resolution-limited downloads now prioritize the requested resolution (with an optional strict H.264/AAC compatibility mode), and hardsub downloads pull the highest available source quality.
- Smarter AI retries: honors server
Retry-Afterheaders on HTTP 503 and short HTTP 429 rate limits, and shows exact wait times on longer rate-limit cooldowns. - Improved AI temporary-file cleanup when confirming removal after hardsub or re-encoding jobs.
- Hardened window security with a strict Content-Security-Policy and navigation/popup guards.
- Layout improvements.
Release v5.0.1
Bug fix and reliability improvement
Release v5.0.0
What’s New
-
The app has been rebranded with a new name and icon.
-
Added a modal for update notifications, offering the option to download or view the release notes.
-
Support for homebrew updates on macOS was added, along with a command copy feature.
-
Windows and Linux users can now update the app directly from within the app itself.
-
Progress tracking has been implemented for update downloads.
Release v4.6.0
What's New
Extract Audio from Local Videos Instantly
- You can now ask the AI: "Extract the audio from this video" or "Get the audio from my downloaded file."
- Instead of re-downloading from the web, the app now pulls the audio track directly from your local video.
Higher Quality Audio by Default
- When downloading or extracting music and podcasts, the app now saves the original, full-quality sound instead of converting it to MP3. This means cleaner sound, smaller file sizes, and instant downloads.
- Still have an older car stereo or MP3 player? Just say "make it an MP3" or "it won't play in my car", and the AI will automatically convert it for you.
Release v4.5.3
Improvements
- Expandable Format Lists: Video format listings (
yt-dlp -F) now attach a collapsible "View full format list" section to the status chip. - Higher Output Limits: Increased tool output buffer to 30,000 characters to prevent truncation on videos with dozens of available formats and codecs.
- Automated Empty Answer Recovery: The AI agent now automatically re-prompts the model if it returns a blank response, preventing empty answer fallback errors.
Bug Fixes
- Fixed Chat Scroll Locking: Resolved an issue where the chat area could get stuck and refuse to scroll downwards in multi-turn conversations.
Release v4.5.2
The download experience for AI Mode has been improved.
Release v4.5.1
Release v4.5.0
What's New
-
Local Media Inspection Tool: The AI assistant can now analyze local audio and video files in your download folder, reporting video/audio codecs, resolution, frame rate, bitrate, channels, and duration.
-
First-Launch Folder Setup: New installations now conveniently prompt you to choose an initial download folder right on first launch.
-
Homebrew Cask Support: macOS users can now install and update the app directly via Homebrew:
Install
brew install --cask shawshank01/tap/yt-downloaderUpdate
git -C "$(brew --repository Shawshank01/tap)" pull
brew upgrade --cask yt-downloaderBug Fixes & Improvements
- Gemini Multi-Turn Reliability: Fixed an
HTTP 400: Missing text in content of type texterror caused by empty text parts during multi-turn tool conversations.
Release v4.4.0
What's New
-
Layout Improvements: New UI with Teal.
-
New App Info & About Modal: Click the icon next to YT-DLP Downloader to view behind-the-scenes details, including the app version, developer information, GitHub repository, bundled tool versions (
yt-dlp,FFmpeg,Deno), and Check for Updates button. -
Live Markdown Streaming: AI chat replies are now progressively rendered in Markdown in real time as the text streams, rather than waiting until the entire response completes.
-
Progress Bar for FFmpeg Tasks: The download/task progress bar now accurately reflects real-time encoding progress during Re-encode as high-quality MP4 and Add Hardsub operations.
-
Optimized API Retries: Refined retry logic for temporary AI server busy (HTTP 503) errors to 1 retry with a 3-second delay, accompanied by an in-chat status countdown.
-
Updated Default Gemini Model: Updated the default Google model to
gemma-4-26b-a4b-it.