Compression fixes, overwrite prompts and security updates:

• Compression no longer removes colour gradients; PDFs with gradients compress less without Ghostscript
• Compression no longer turns pages to follow the direction of their text
• Compression keeps the OCR text layer and, in most cases, links, form values, attachments and metadata
• On Linux, compression runs Ghostscript with the system's own libraries
• Converter and Import ask before replacing an existing file
• AUR pdfapps-bin: no more checksum failure on update, and missing dependencies declared
• AppImage, winget, Copr, Flatpak and the GitHub .msix are retired; use the Snap, the AUR or the tar.gz
• Security: pypdf 6.21.2 or later, closing three CVEs; Pillow 13 on Python 3.11 or later; Ghostscript 10.08.0 in the Windows installer

OCR and compression fixes, and security updates:

• OCR runs in the language you pick, instead of the first installed one
• The OCR language list shows only installed languages and keeps your choice
• Chinese and Dutch appear in the OCR language list under their translated names
• Running OCR no longer closes the app when an old or unrecognised Tesseract is installed
• OCR and compression no longer open a console window on Windows
• OCR fixed in the Snap package, which could not find its bundled language files
• Compression in the Snap package now runs its stronger stage, which it skipped before
• Compressing some damaged PDFs on Linux no longer hangs for about two minutes
• Starting from a terminal no longer prints a deprecation notice
• The Help button opens the current documentation page
• Security: pypdf 6.19.0, closing three advisories present in 1.15.0, and urllib3 2.8.0, closing three

Password fixes and a stronger encryption stack:

• Correct passwords with accented or non-English characters are no longer rejected by the tools after unlocking
• The password spelling that worked is reused across the viewer and every tool
• A wrong password now shows a clear translated message
• AES-256 encryption fixed in the Flatpak build, where a missing cryptography module made it fail
• Security: pypdf 6.16.2 and cryptography 50.0.1, closing seven advisories

OCR language packs installed through Homebrew are now detected on Apple Silicon Macs.

Viewer stability, legacy-encoding imports, and security updates:

• The shared document handle stays valid after a failed note deletion, and saving no longer races with rendering
• In-document search is debounced, so typing a query no longer blocks the interface
• Text files in Windows-1252 and Latin-1 encodings now import without errors
• Security updates to pillow, pymupdf, pypdf, and qtawesome

Page thumbnails sidebar in the viewer:

• A Pages tab shows a live thumbnail of every page; click one to jump to it
• Thumbnails render on demand and stay sharp on high-DPI displays
• A Contents tab shows the document outline when a PDF has bookmarks
• Password-protected documents are verified correctly when opened for viewing

Editor audit fixes, atomic saves, and streaming OCR:

• Encryption is preserved on save and annotation deletions persist
• Atomic writes with restore-on-error added across the tools
• Streaming OCR writer lowers peak memory use
• PDF to DOCX conversion is now layout-aware

Five new import formats and four new export formats:

• Import DOCX, PPTX, XLSX, HTML, and EPUB to PDF
• Convert PDF to PPTX, XLSX, HTML, and EPUB

New features and fixes:

• Global keyboard shortcuts (Ctrl+O/S/P/W)
• Multi-select PDFs in file dialog and drag & drop
• Canvas background adapts to light/dark theme
• Theme-aware dialogs in light mode

Digital signature tool, continuous scroll editor, and security hardening.