KDE Plasma 6.6.5 is mostly a bugfix release, improving performance for NVIDIA GPU users who experience issues introduced by the NVIDIA 595 graphics driver, and improving the login and lock screen experience by addressing a Plasma Login Manager crash when connecting and disconnecting multiple monitors and a startup failure when using certain graphics hardware.
Coming two months after LibreOffice 25.8.6, the LibreOffice 25.8.7 release is packed with more fixes to address various bugs, crashes, and other annoyances reported by users in an attempt to improve the overall stability and reliability of this popular open-source, free, and cross-platform office suite.
Fwupd 2.1.3 is here about two weeks after fwupd 2.1.2 with support for SHIFT6mq and SHIFTphone 8 modular smartphones, support for Redfish bearer token authentication, support for several XMC SPI chips, and support for parsing JCat files in libfwupd without using libjcat.
Coming about a month after GStreamer 1.28.2, the GStreamer 1.28.3 release introduces hardware-accelerated H.265 encoding support for NXP i.MX 8M Plus SoCs to the webrtcsink element, a leaky mode to the dataqueue-based elements, and fallback-source and enable-dummy properties to the fallbacksrc element.
Debian’s Paul Gevers reports that the decision was taken for Debian to ship reproducible packages, which means that if you take the same source code, the same build instructions, and the same environment, you can build a binary package that’s bit-for-bit identical every single time. This will be a requirement in Debian 14 “Forky,” and non-reproducible packages will be blocked.
Based on and fully compatible with the Debian 13.4 “Trixie” repositories, SparkyLinux 8.3 ships with Linux 6.12.86 LTS as the default kernel on the live system, with the latest Linux 6.12.87 LTS already in the repos, as well as support for Linux kernels 7.0.6, 6.18.29 LTS, and 6.6.125 LTS kernels, which users can install from the official SparkyLinux repositories.
The MX Linux 25.2 release is the second update in the MX Linux 25 “Infinity” series and promises to ship with a much-improved installer that now features a new text mode, allowing you to install MX Linux in a terminal emulator by running sudo minstall --tui or a text console by running minstall-launcher.
I want to thank everyone who sent us donations; your generosity is greatly appreciated. I also want to thank all of you for your continued support by commenting, liking, sharing, and boosting the articles, following us on social media, and, last but not least, sending us feedback.
Audacious 4.6 promises a new File Browser plugin, which will be available for both GTK and Qt interfaces, a macOS Now Playing plugin, support for exporting playlists via command line with audtool, support for playing Musepack SV8 files, and support for all AIFF extensions and MIME types.
Shelly 2.2.4 is the fourth maintenance update in the Shelly 2.2 series, but an important one that introduces smarter fuzzy search across every package list, a refactor of the built-in search feature with package group search and sortable search columns, support for build dates in package details, and fingerprint authentication support.
Milk-V has introduced the Jupiter2, a compact RISC-V single-board computer based on the SpacemiT Key Stone K3 processor. Similar to the recently announced Sipeed K3 Pico-ITX platform, the board combines eight X100 RISC-V CPU cores with an eight-core A100 AI subsystem rated for up to 60 TOPS, LPDDR5 memory, and high-speed networking interfaces including 10GbE SFP+.
Sipeed has opened pre-orders for the SpacemiT K3 CoM260 Developer Kit and K3 Pico-ITX, two RISC-V AI computing platforms based on the SpacemiT Key Stone K3 processor. The systems combine eight X100 RISC-V CPU cores with eight A100 AI-oriented compute cores delivering up to 60 TOPS of AI performance for edge AI and embedded workloads.
SpacemiT’s Key Stone K3 is a high-performance RISC-V SoC designed for AI and edge computing applications. The processor combines eight X100 64-bit RISC-V CPU cores with eight A100 AI-oriented compute cores, along with multimedia, networking, and high-speed I/O support targeting edge and embedded AI workloads.
