Anyone who wants a break from Linux is usually told to try another distro. That advice, while good, leaves out those who want a break from Linux itself.

Every system listed below is open source and not powered by Linux. Each has an ISO or installer you can boot today, and many of the picks can even run Linux applications.

Think of it as a list of operating systems that are friends with Tux🐧 but not friendly enough to call him inside for a cup of tea.

1. Vinix

If you own an Apple Silicon MacBook and want a system that is neither macOS nor a Linux distribution, Vinix is aimed at you. Only the original M1 is supported today, and Alexander Medvednikov, the author of V, maintains the project.

The kernel is monolithic and written mostly in V, while part of it is C and assembly. Because memory is managed by hand, no garbage collector can interrupt a system call, and the security model borrows certain concepts from OpenBSD.

On ARM64, it runs Alpine Linux binaries natively, and pkg installs packages from Alpine's aarch64 repositories. ISOs for both x86-64 and ARM64 are available for virtual machine use.

2. Asterina NixOS

Trying Asterinas NixOS means setting up a NixOS system and finding a Rust kernel where Linux should be. Asterinas is that kernel, and because it implements Linux's system call interface, apps for the Linux platform run on it.

The distribution is configured through Nix like regular NixOS, and unsafe Rust is allowed only inside a small framework called OSTD, while the rest of the kernel, along with its drivers, is safe Rust.

x86_64 is the fully supported platform, and ARM64 and RISC-V sit one tier below it. The project says the distro is not ready for production use and aims the installer ISO on GitHub at early adopters and enthusiasts.

3. FreeBSD

FreeBSD is a complete operating system derived from BSD, the Unix version developed at the University of California, Berkeley. A large community maintains it, and its license lets companies keep their own changes private.

ZFS, jails, and a hypervisor sit at the center of what it offers. OpenZFS is built into the base system, and boot environments let you roll back a bad update by snapshotting the system first.

Its recent release, FreeBSD 15.1, came out earlier this year, and the OS runs on hardware platforms like AMD64, AArch64, Armv7, PowerPC, and RISC-V.

4. GhostBSD

GhostBSD is FreeBSD with the desktop work already done. The project pairs a FreeBSD base with the MATE desktop and offers a set of applications that make users feel familiar with using it.

The name started as shorthand for GNOME hosted by FreeBSD. MATE, the desktop it ships today, is a continuation of GNOME 2, and the old name still fits.

Updates arrive on a slow-rolling schedule, and the project says it values a stable cycle over chasing the newest software the way some Linux distros do. The current release, 26.1-R15.0p2, came out in April on a FreeBSD 15.0 base.

Also good to know is that Linux binary compatibility comes preconfigured, and an installer built on OpenZFS keeps setup approachable for newcomers.

5. Haiku

Haiku is an open source operating system inspired by BeOS that is aimed at personal computing. The same team writes the kernel and the desktop applications, with the project crediting that approach for the system's consistent interface.

Under that interface sits a kernel built for responsiveness, a system that is multithreaded throughout, and a BFS file system whose indexed metadata lets Tracker, the file manager, run database-style queries across a drive.

Software installs are handled via packages that mount into the system, and earlier states are kept for rolling back botched updates.

6. ReactOS

ReactOS comes with a warning label attached because the OS is still considered alpha-level software, and the download page promises nothing about stability, the safety of your files, or whether it would work on your hardware.

Behind the warning is a full operating system with a Windows-style desktop that the project says was written from scratch, while most of its user-mode code is based on Wine, the ubiquitous compatibility layer.

By default, it installs only onto a FAT16 or FAT32 partition, and Btrfs support is experimental.

A suggestion

Try these operating systems in a virtual machine before switching. I say that because it’s a good habit to test out such non-standard pieces of kit before committing to them on your hardware.

First see what works for you. Then see if your workflow and usage habits align with what the OS offers. Once you settle on one, keep it updated so you never miss critical patches.

Air-gapped setups are better protected from such risks, of course.

Valve's October 15 Steam Client Beta changes how the Linux client is built and run. It moves the client to a 64-bit build and runs it inside a Steam Runtime container. Alongside that, a fix for the updater window failing to show up is included, as is the new Achievement Sets feature.

According to Valve, the client has run in an older setup that layers Steam Runtime libraries over the ones on the host system. The runtime documentation adds that this only works while the host is newer than the runtime.

Games on Steam can already run in containers, which give an application its own set of libraries rather than blending them with the host's. The tool behind those containers, pressure-vessel, still takes graphics drivers from the host system.

Valve says running the client in a container should make it behave more consistently across distros. Its games already run on the same technology.

The release notes for the beta client do not name the container, but its runtime release notes tie the 64-bit client to one called steamrt3c. The README describes it as a variant of the sniper runtime made for use inside the Steam client.

The groundwork for this can be traced back to March, when Valve put the container and the early 64-bit client out as the Linux SteamRT3 Beta.

And don't assume that opting for a 64-bit client removes the need for 32-bit libraries.

The requirements for the beta client still call for both 32-bit and 64-bit versions of the required libraries and graphics drivers on x86_64 systems.

They also list 64-bit-only setups as unsupported. The steamrt3c README from earlier notes that its containers ship i386 libraries for compatibility with older software.

Test it now

I switched to the beta on Ubuntu 26.04 LTS, and it only took me four steps. If Steam is not already on your system, you need to get the official DEB package for the client and install it.

Next, follow these steps. 👇

1. Open Steam, click Steam at the top left, and choose Settings.
2. Go into the Interface tab and find Client Beta Participation.
3. Open the dropdown and select Steam Beta Update.
4. Click Restart Now when Steam asks you to restart.

That is all it takes, and Steam is now on the beta channel. Valve welcomes bug reports from testers, so if you run into an issue, use the "Report a bug" button in the Interface page.

Suggested Read 📖: Epic Games is looking to entice Linux gamers with its anti-cheat move.

The title says that because, until now, the core utilities in Ubuntu were not fully Rust. cp, mv, and⁣ rm had stayed on the GNU versions over compatibility issues. They have since moved to the Rust-based uutils implementation, taking the distro even further in its oxidation process.

Ubuntu 26.10, codenamed "Stonking Stingray", is an interim release, which means it brings newer software than the earlier LTS while only getting nine months of support, ending in July 2027.

Although our features roundup covered a lot of ground, there were some changes that it missed. 😅

⭐ Ubuntu 26.10: What's New?

Well, we start with the kernel powering this interim release, Linux 7.3, which offers support for recent Snapdragon X2 Elite laptops along with several older Snapdragon X Elite and X Plus machines from Lenovo, HP, and Acer.

Gamers will be glad to know that the first chunk of Valve engineer Natalie Vock’s VRAM overcommit work has also landed in this kernel. Games on graphics cards with too little video memory should run better, as some of the data can now be offloaded to regular system RAM.

You already know that this Linux release is shaping up to be one of the most performance-focused releases in years.

Ubuntu is also moving to the NVIDIA 615 driver series, which replaces 595 and 610 and ships open kernel modules only, supporting Turing and newer GPUs.

The rustification we talked about earlier also extends to OpenPGP, where Sequoia PGP, a Rust-based implementation built by former GnuPG developers, has landed in the main archive. Its sq and sqv tools now sit alongside GnuPG's gpg and gpgv.

For the desktop environment, Ubuntu 26.10 has GNOME 51, with its range of upgrades such as a fingerprint manager in the Users panel, DNS search domains in the Network settings, and a count badge in Files when you drag multiple items.

There's also the Ubuntu Tiling Assistant that shows a Windows-11-like interface for the built-in snappable window layouts.

If you were hoping for a redesigned installer and first-boot onboarding flow, those are planned to come for the next long-term support release, Ubuntu 28.04 LTS.

🛠️ Other Changes & Improvements

These are some of the other notable changes in this release:

• OpenSSL 4 brings Encrypted Client Hello (ECH) and better QUIC support.
• TPM-backed full-disk encryption now works without a hardware root of trust if a PIN or passphrase is set.
• The RISC-V platform gets official RVA23 support for the SpacemiT K3 boards and SiFive's BigSky platform.
• Twelve new wallpapers that were picked across the Mascot, Photography, and Digital/Abstract categories.
• The toolchain upgrades consist of GCC 15.2.0, LLVM 22.1.8, Python 3.14.7, Rust 1.97.1, Go 1.27, and OpenJDK 25.0.4.

📥 Download Ubuntu 26.10

ISOs for this interim release are available for AMD64, ARM64, RISC-V, and others from the official website; go for the "Desktop" ISOs if you are a regular user. Specific flavors like Server, Netboot, and WSL can be fetched from the releases portal.

The European Commission's proposed Cloud and AI Development Act, or CADA, looks to lay out four tiers of requirements for cloud services where public bodies would pick a tier for each service they buy, based on the risk involved.

The upper tiers demand tighter control over who owns a provider and who controls its software. ECIPE, a Brussels-based think tank, put a 10-year cost estimate on those rules via a research report this week.

It argues the Commission's impact assessment misses an important part of the costs, mainly one-off software rewriting.

CCIA, or the Computer & Communications Industry Association, funded the study. The association is a tech trade group with members like Amazon, Google, Intel, Cloudflare, Apple, and Meta.

Three scenarios, one estimate

ECIPE modeled three scenarios for the period between 2027 and 2036. The narrow one comes to €35.6 billion, the middle one to €115 billion, and the broad one to €233 billion.

Though none of them are being presented as a forecast. The difference is how many workloads end up in the two strictest tiers and how far the rules reach into private companies.

The middle case puts 20% of affected workloads in those tiers, and it also applies the rules to a quarter of cloud use in essential private sectors like energy and banking.

About €33.3 billion lands on public budgets. That works out to roughly €159 for each person part of the working-age workforce.

Around €80 billion, which is about 70% of the total, would fall on private companies. That only happens if the Commission uses Article 31 of the CADA to extend the rules to them.

Where the Money Goes and Who Feels It

About 69% of the total is paid for moving to new providers and rewriting software. Higher ongoing cloud prices account for roughly 30%. Postponed digital projects make up the last 2%.

This shows that moving is not a simple fork it and forget it job. There are plenty of cases out there where an app built around one provider's services may need parts of it rebuilt for another.

The EU average in the middle case is about €548 over 10 years for each person in the workforce. Germany sits at €396. Luxembourg is highest at €1,711, though its workforce count excludes cross-border commuters.

Since Germany, France, Spain, and Italy already apply national cloud rules to some government bodies, they face a lower blow to their costs in the study's model.

Smaller countries that have little homegrown supply at the strictest tiers may end up buying from bigger EU markets at higher prices. ECIPE cautions against reading the gaps as proof of any one cause.

What the study leaves out

The paper calls itself independent and says the views are the author's own. It puts no value on security or resilience benefits and says this is not a full cost-benefit analysis.

The Commission calls over-reliance on non-EU cloud providers a significant risk to Europe’s digital autonomy and resilience.

CCIA wants narrower rules. Its release asks for the sovereignty requirements to be kept for sensitive workloads and for technical safeguards to replace rules based on where a company comes from.

AI's role in kernel work is looking more apparent now. Linus Torvalds' use of Claude Code has tripled read speeds for concurrent file access, while Sasha Levin proposed an agents directory to house AI skills for kernel work.

And that's just the way AI is progressing, for good or bad. Last week's most popular news was an open source project that used AI to reverse engineer Adobe's Photoshop and other tools. Now the same project has used AI to 'clone' Microsoft Office.

These days, 9 out of 10 new open source software I come across is created with AI assistance. My inbox is flodded with new software review requests. One glance and I can 'feel' that both software and the coverage request email used AI.

So, like Linus Torvalds, I have made my peace with it. Until now, I was trying not to recommend open source tools that were openly developed with help of Claude or other AI tools. I don't see the point of it anymore now. If an open source software is good and I feel that readers would find it useful, I will feature it regardless. You can then decide on your own if you want to use it or not.

And this is also the time to get moving towards local AI. Where AI and models live in your system locally instead of a cloud. Sure, it won't match the performance of Claude and other frontier models, but some smaller tasks can still be achieved with local AI. Ubuntu's AI-based voice input tool Myna is a proof of that.

Do subscribe to the Local AI Weekly newsletter if you want to stay updated and use the AI on your system in true open source spirit.

This edition of FOSS Weekly is supported by PrivacyNotes, an open source, zero knowledge note management app.

Your notes, tasks, password, secrets. Keep them organized with open source tool, PrivacyNotes. Store them locally or encrypt them to a Swiss cloud. Start free, import the notes you already have, and see if it fits. If you want more storage with Pro later, it’s a one-time payment, not a subscription.

Try PrivacyNotes

📰 Linux news that matters

Ubuntu 26.10 is going to be released today. While we wait for Ubuntu 26.10 ISOs to appear, the next interim release already has a codename, "Tenacious T-Rex," marking the first time Ubuntu has ever used a dinosaur-themed mascot.

ONLYOFFICE Docs 10.0 brings AI assistance to every editor for text summarization and formula drafting, while adding collaboration features, spreadsheet speed improvements, and modular apps for forms and file management.

Strata is a new keyboard-first file manager for Omarchy and Arch that just hit 1.0, offering column view browsing, vi-key navigation, and a clever feature that remembers each folder’s sort order and icon size.

LibreOffice has come to the rescue after Microsoft's discontinuation of Publisher for 365 users. Using the Document Liberation Project’s reverse-engineered library, users can now open PUB files and migrate them to open standard formats.

Mozilla is targeting the commercial market with Firefox Enterprise, a product that would let organizations manage Firefox installations across employee devices through a self-hosted console.

Linux Mint has moved to an annual release cycle, with the next one, Linux Mint 27 "Adrien" arriving on Christmas. There won't be a Mint 23.x

The CachyOS team recently announced that Framework is increasing its support for the project, going from $250 per month to $850, taking the total figure up to $10,200 a year.

Open source boot firmware LibreBoot has a new release supporting more hardware now.

🧠 What We’re Thinking About

Six open source vendors have launched Open Productivity Software to expand openDesk from government use into the enterprise market.

🧮 Linux Tips, Tutorials, and Learnings

My Computer is a Shell extension that restores the unified "Computer" view that GNOME dropped from Files (Nautilus) in the GNOME 47 release. It displays drives, network locations, and removable media in one place rather than them being scattered across the sidebar.

Arch users run pacman -Syu for updates because it prevents dependency mismatches when the local package database is stale. Running just -S on an outdated system can leave you with incompatible package versions.

👷 Homelab and Hardware Corner

Pebble completed the full open source release of software, hardware schematics, and mobile apps in 2025, paving the way for its resurrection. Now the Pebble Round 2 smartwatch has entered production and begun shipping to preorders.

Zima, brand behind ZimaOS, ZimaBoard, and ZimaCube, is organizing a self-hosting event on Reddit where you can win homelab gadgets and AdGuard and PhotoPrism licenses just by sharing your self-hosting story. If you already self-host, why not try your luck? Or, at least explore how other people are self-hosting.

✨ Apps and Projects Highlights

BentoPDF is a privacy-first, open source PDF toolkit that processes files locally in your browser instead of uploading to servers.

A new open source GIS tool is here; it lets you chat with an AI Assistant about map data and geographic workflows.

📽️ Videos for You

For people who are working extensively with git, this terminal-based git client is worth a try.

💡 Quick Handy Tip

In the Nautilus file manager, you can carry out a search in the current folder using Ctrl+F or search everywhere using Shift+Ctrl+F. When you find the item you were looking for, you can use the shortcut Ctrl+Alt+O to open its location.

Going further, you can also use the shortcut Ctrl+. to open a terminal at that location.

🎋 Fun in the FOSSverse

Call yourself a true Arch Linux fan? Are you bold enough to prove it?

A reality check for big names like Adobe. 🤭

🗓️ Tech Trivia: Long before IT became the industry we know today, British computing professionals were organizing themselves. On October 14, 1957, the British Computer Society was formally incorporated following a merger of two computing groups.

🧑‍🤝‍🧑 From the Community: A very interesting thread surrounding NixOS and Nix has popped up, featuring replies from supporters and skeptics.

Linux gamers have gotten used to hearing that yet another multiplayer title is cutting off the platform due to a variety of reasons, though most of their arguments usually sound lazy to me.

Take Ubisoft, for instance. They recently stopped For Honor from being playable on Linux and the Steam Deck, saying that they could not meaningfully deploy protection for the platform.

Now we have Epic Games swaying the other way, rebuilding its Linux anti-cheat under a new name while also offering a preview release for the long-requested native Epic Games Launcher.

Anti-cheat on Linux isn't some pipe dream

Easy Anti-Cheat has supported native Linux games since 2014 and Windows games running through Wine or Proton since 2021. Epic is now rebuilding that Linux implementation into EAC Linux v2, aiming to get much closer to Windows by leaning on security features Linux already has.

On Windows, EAC relies on a kernel driver that is active only during play. On Linux, EAC Linux v2 won’t have such a module acting in one of the most sensitive parts of the system.

That matters because kernel-level anti-cheat is a common sticking point for Linux gamers.

EAC Linux v2 runs a systemd service that stays active just for the length of a protected game session, alongside eBPF programs written by Epic that watch for tampering and cheats.

For players, this means a stricter boot setup, so Secure Boot and TPM 2.0 need to be on, the kernel has to be signed with a key Epic recognizes, and several other boot and kernel settings come into play.

EAC checks all of this only when a protected game launches, and failing the check brings up an error, not a ban. Players can also keep a separate boot entry for protected games and run their own keys elsewhere.

Testing the waters

Fortnite will be the first game to test EAC Linux v2, with plans to open it to other developers once the system is validated. Developers with games already on EAC for Linux can stay on the older version or switch.

Getting Fortnite itself onto Linux depends on distro support. It will arrive through the new launcher (shown below) once enough maintainers meet the security requirements.

The preview officially supports Ubuntu 22.04 LTS or newer, Fedora 43 or newer, and Debian 12 or newer, while other distros may work but are not officially tested.

The launcher itself runs natively, but many games are Windows-only, so they may need a compatibility layer like Wine or Proton to run.

The community needs to step in

Epic is reaching out to Linux community leaders and inviting distro maintainers to get in touch since protected games will only launch on systems booted with keys from a trusted list.

It is maintained in-house for now, with the goal of going open source sometime down the line.

They already seem to be in touch with Fedora, as a panel scheduled for later this year names Fedora Project Leader Jef Spaleta as a co-host, with the talk centered on enabling fair competitive gaming on Linux.

SteamOS is the next obvious name to bring up, but the Steam Deck doesn't meet the requirements yet. Talks with Valve are ongoing, with the aim of supporting the handheld later.

While it's still early, this is good news for gamers on Linux, as this is a platform that keeps getting shown the proverbial middle finger when it comes to gaming.

Suggested Read 📖: Last month, I spent some time with GeForce NOW on Linux to see if it's gameable.

Collabora Productivity, Element, Nordeck, Open-Xchange, Univention, and XWiki have put their names behind a new Berlin-registered company called Open Productivity Software GmbH.

Leading the firm are the two managing directors, Dr. Gerald Pfeifer and Peter Ganten, who are tasked with making it all work. Pfeifer held leadership roles at SUSE before taking the job, while Ganten is the CEO of Univention.

If you were wondering what openDesk is, it is a suite of office and collaboration tools made up of open source apps from different vendors. The new company is meant to market it to critical infrastructure operators and businesses internationally.

The project itself is an offering by the German ZenDiS, which has focused on public administration so far and launched a private-sector partner program for the suite last month.

Two of the founding members, XWiki and Open-Xchange, are also part of the Euro-Office initiative, the ONLYOFFICE-based document project we covered earlier this year.

What's inside openDesk?

Each founder supplies a piece of the suite. Here's a brief summary of who does what:

• Collabora Productivity handles document editing with its online office suite.
• Element brings real-time communications on top of Matrix.
• Nordeck adds communication tools built around Jitsi and its NeoBoard whiteboard.
• Open-Xchange covers email, calendars, and collaboration.
• Univention looks after identity and access management.
• XWiki supplies the wiki system for knowledge management.

Nextcloud and OpenProject are part of openDesk too, handling files and project planning, but they are not among the founders. The company also names Microsoft 365 and Google Workspace as its competition, pitching openDesk as a flexible alternative that gives users full control of their data.

How the partner model works

Customers are meant to deal with a single contact per project, who covers everything from consulting and migration to day-to-day operations. Hosting on a customer's own infrastructure or in the cloud is still something that's being worked on.

Three launch partners are backing the newly formed company. Adfinis is a Swiss-headquartered open source firm, the msg group has served critical infrastructure operators, while the French firm Smile brings integration expertise.

Open Productivity has not disclosed whether more partners will be onboarded following the launch.

Speaking during the launch, Pamela Krosta-Hartl, the Managing Director of ZenDiS, added that:

We welcome the initiative of several openDesk vendors to support openDesk’s path into the enterprise market through Open Productivity. Open Productivity enables the distributors and systems integrators that ZenDiS is now addressing with its new partner program.

You can go through the press release to learn more about their intentions with this joint venture.

Suggested Read 📖: openEuler wants Linux to treat AI clusters like a standalone machine.

Gaël Duval has already put out two beta releases of ElixirOffice, a desktop app and a self-hosted server version, and he is asking early testers for feedback.

His wife, a teacher, has been facing a predicament many people usually face when using open source office suites. For her, the editors offered at school felt unfamiliar, the menu items and shortcuts felt out of place, and not to mention colleagues' documents also looked off.

ElixirOffice follows the Microsoft Office-style ribbon, menus, and shortcuts, and it opens OOXML document formats as well as their OpenDocument counterparts. It targets Linux, Windows, and macOS, with a server version aimed at teams.

This is what's in the beta

Both of those versions use the AGPL-3.0 license, where the desktop app works offline, while the server version runs through Docker on a machine you host.

ElixirOffice offers substitutes for the usual Microsoft Office apps. The document editor takes the place of Word, the spreadsheet editor of Excel, and the presentation editor of PowerPoint.

Each one opens the Microsoft format and its OpenDocument equivalent, so .docx and .odt files go to the document editor, .xlsx and .ods to the spreadsheet editor, and .pptx and .odp to the presentation editor.

The server version also lets you connect Murena Workspace and Nextcloud accounts through an optional start page, which stays disabled unless you switch it on, and the desktop editor has no support for accounts, mail merge, co-editing, version history, or password-protected files.

On Linux, the desktop beta comes as an AppImage and a DEB for Debian and Ubuntu, whereas Windows gets an installer and macOS a disk image for Apple Silicon devices only.

How did this come about?

Gaël started and supervised the project, while large language models handled the coding via AI agents, with tests and reviews being run on every change. ElixirOffice also has optional AI features that stay off until you connect a model of your own.

To keep layouts close to Word without Microsoft fonts, it ships free metric-compatible ones, plus Marelle, a font commissioned by the French Ministry of Education. The README file on the project's repo also claims that no telemetry is collected and sent.

Its roots

ElixirOffice is not written from scratch. It sits on top of Euro-Office, which itself comes from ONLYOFFICE, and the project's changes live inside the forks of the Euro-Office components.

The original copyright notices and license files are kept in place, and the project states that it is not affiliated with Ascensio System, the company behind ONLYOFFICE, or the Euro-Office project.

To provide you with some context, Euro-Office is an editor that European companies, led by Nextcloud and IONOS, are building to run inside other platforms. It launched on March 27, 2026, and shipped its first stable release on June 9.

Around that time, ONLYOFFICE had raised licensing and attribution concerns about Euro-Office and suspended its partnership with Nextcloud, which disputed the claims.

Who's Gaël Duval?

Gaël is best known as the creator of Mandrake Linux, a distro that later became Mandriva. He left the company in 2006 and went on to start Ulteo.

Later, he began work on eelo, which is now called /e/OS. It is a de-Googled, Android-based mobile OS from the nonprofit e Foundation, where Gaël is listed among the founding board members.

He is also the CEO of Murena, the company that sells phones running /e/OS and runs the Murena Workspace cloud service, which was previously called eCloud.

Could it end up in Murena Workspace?

Nothing in the project's documents or Gaël's announcement says Murena plans to offer ElixirOffice as part of Murena Workspace, so any such move can only be called speculation at this point.

The server version connecting to Murena Workspace accounts is a promising sign, but that doesn't really mean anything right now.

Back in 2022, ONLYOFFICE said that Murena Cloud used its editors, and the e Foundation's landing page for the service says Murena Workspace is powered by ONLYOFFICE too.

So my thinking is that, if ElixirOffice does mature to the level of ONLYOFFICE, there's not much stopping it from being adopted as the in-house base for Murena Workspace.

For now, the source code and the releases for ElixirOffice can both be found on GitHub. 👇

Suggested Read 📖: Seeing that we are on the topic, ONLYOFFICE Docs 10.0 is out now with many upgrades.

This week, AI went to the "office". Linus Torvalds says he actually likes using AI, and his new kernel branch credits Claude Code for the code, the tests and even a draft of the commit message. Office suites are being rebuilt so agents can edit your documents. And Big Tech wants its agents inside every work app you use, including the office app.

The question I keep asking is whose desk the agent sits on. If it lives in someone else's cloud, so does your work. That's why the local and self-hosted options matter.

🧪 On my desk: openKylin 3

I've been testing openKylin 3.0. It ships with Kylin bot, an AI assistant that can change desktop settings for you. With openKylin's own hosted model, I could successfully make the agent change the system settings.

But my local and frontier models did not fare as well. When assigned a task, my Qwen model with Ollama replied that it changed system settings when it was not changed in reality.

I will be sharing all this in a video later this week. Stay tuned for that.

🔍 Discover AI tools

BetterOffice is an Apache-2.0 office suite you embed in your own web app, with DOCX, XLSX and PPTX editors built on Rust and WebAssembly. The interesting part is that an AI agent edits as a peer, and its changes show up as tracked changes you accept or reject.

ArtCraft, the folks behind those Adobe clones, has moved on to Microsoft Office. WordCraft, GridCraft and DeckCraft are Rust clones of Word, Excel and PowerPoint, built with Claude. The READMEs claim high parity with Office, but the numbers are self-reported and the apps are pre-alpha. I'll believe it when it is tested in real world, with real use cases.

Razorpay open sourced slashdocs, a self-hosted place where agents publish HTML reports and your team comments on them. The agent reads the comments and pushes the next version. It's MIT licensed and ships as a single Go binary.

Raven calls itself a "harness of harnesses." It splits a task into pieces and hands them to specialist agents, including Claude Code, Codex, OpenCode and Hermes. It's Apache-2.0 and pre-alpha (again). Sorry but so many new tools in alpha stage these days. But that's the joy of discovering new tools, right?

📡 Open Model News

Two big announcements this week, not that you can download them (yet).

Mistral Large 4, nicknamed "Le Chonk," is a 1-trillion-parameter multimodal MoE with roughly 50B parameters active. It's API-only in preview, and Mistral promises the weights by the end of October. No license has been announced yet.

Reflection AI's Beam is a 501B model with 23B active, aimed at coding and agentic work. The weights are due later this month under Apache 2.0. Reflection's benchmark claims haven't been independently verified yet.

Neither will fit on our regular machine anyway. If you want something small for coding agents, JetBrains released Mellum2.1 Thinking, a 12B model with about 2.5B parameters active.

👀 Big Tech Watch

At its Windows event, Microsoft pitched agents that run on your device when possible and in the cloud when not. Its MAI-Code-1.1-Flash runs locally as a 3-bit build with 256K context. Local calls are free, but the recommended RAM is 120GB or more (LOL).

Meanwhile, Google's new Gemini agent works across Workspace, Microsoft 365 and Slack for hours or days at a time. OpenAI launched always-on Dots agents in ChatGPT. Meta plugged its Muse agent into Slack, Shopify and QuickBooks. Everyone wants a seat at your desk.

🗂 AI Jargon: Tool calling

Remember Kylin bot claiming it changed the system setting? That was a tool calling failure.

A model can't touch your computer on its own. It can only produce text. Tool calling is how it asks the app to act: the app tells the model which tools exist, like set_volume, and the model replies with a structured request to use one. The app runs the tool and sends the result back.

Smaller models are often bad at this. They describe the action, or skip the call entirely and just say it's done. That's why tool support matters when you pick a model for an agent.

😂 Meme

An old meme that never really gets old.

⚡ Quick Tip: Check if your Ollama model can call tools

A quick check before you plug a local model into an agent like Hermes.

Run ollama show <model-name> and look at the Capabilities section. If tools isn't listed, the model wasn't trained for tool calling, and your agent will struggle no matter how you configure it.

Even if tools shows up, test it with a harmless action and verify the result yourself.

See you next week.

In the project's September monthly update, Linux Mint lead, Clement Lefebvre (clem), says that the distro's version number now reflects the upcoming year and that point releases aren't a thing anymore.

Releases will come once a year, so 28 follows this one, and users upgrade from the current year's to the next. This ought to make it simpler for the many beginners who use Linux Mint as a gateway to the Linux platform.

The codenames change too, as the girls' names Mint has used since 2006 give way to boys' names. Linux Mint 27 "Adrien" is the first release to follow these new conventions, though it did not start off that way.

Where did LM 23 go?

You see, the devs had no name or version number ready for the next release when work on it began earlier this year, so they picked temporary ones. This way, "Alfa" became the codename and "23" the version.

That was because the release strategy was unclear. The cycle length, the handling of minor releases, and the question of alpha releases were all undecided, and the strategy would also dictate how releases got numbered and named.

It all goes back to the January update, when Clem said shipping a new release twice a year while also looking after LMDE was leaving them with less time to actually build features.

By March, the team had locked in a longer development cycle and a Christmas 2026 target for the next release. The move we have been talking about so far marks the end of that long decision-making process.

Clem also points out that using the girls' names for releases all these years, from Ada to Zena, had made them run out of alphabet just as the schedule changed, which he found a funny coincidence.

He says the project has never spent a full year on one release before, and while the move away from six-month cycles was a tough decision, it is working out.

A year of development

The longer cycle gives the team room to attempt projects that would once have felt out of reach. The Clockenstein project, which gave Mint a calendar app in the August update, now has an alarm app called Clocks too.

Alerts from Clocks run through the Clockenstein daemon and agent that already handle the calendar's notifications, so they go off even with the app shut. It handles repeating alarms and lets you pick your own sounds, while its look and behavior are borrowed from Android.

Three more projects, Superlight, Unfold, and Velocitty, are also getting their finishing touches. The next update blog will feature them.

If you need a refresher of the development work carried out so far, then you can read about Linux Mint's kernel removal feature and how Wayland support has been achieved.

A file manager is one of the most basic tools a Linux distribution comes with, and desktop environments ship their own, such as Nautilus on GNOME, Dolphin on KDE Plasma, and Nemo on Cinnamon.

It is where people browse folders, move and rename files, preview what is inside them, and clean up any cluttered directories. Thanks to the platform's openness, you can also install more than one file manager on your system and use whichever you prefer.

A new one that has made an appearance recently is Strata, which is tailored for Omarchy and, to some extent, Arch Linux. Other glibc-based distributions might be able to run it, provided they meet its library requirements.

This file manager is written in Rust with GTK4 and released under the MIT license, with its lead developer, Pierre, announcing its first stable release a few hours ago. He claims that in 45 days, the project has picked up over 30 contributors and 700+ commits, with ~300K lines of Rust now in the codebase.

Strata is Here

It can show folders side by side, so opening one reveals its contents in a new pane to the right while its parents stay visible, like the column view in macOS Finder. Icon and list views are available too, and being a keyboard-first app, it supports navigation with the arrow keys or the vi-style h, j, k, and l keys.

Ctrl+K opens a fuzzy search over your Home folder and every mounted local drive, which matches file names and paths but not file contents. The 1.0 release adds exclusions in Settings for skipping folders like node_modules or a single directory.

Tabs are another new addition that let any folder be opened in a new tab, with keyboard support for switching/reordering them, and previously open tabs being restored after a restart.

Strata also stores sort order and icon size per folder, where one folder can be sorted by date with large icons while another stays alphabetical with small ones, and each opens that way the next time.

Pressing Space toggles a preview panel for images, camera RAW files, PDFs, source code, audio, and video.

It includes a diverse range of themes too, and an in-app editor lets you create your own and save them as TOML files. Code and config files also get language-specific icons based on Catppuccin, in brand colors or the theme's accent.

Install it

Strata has an interactive installer for Omarchy and Arch Linux that runs from a single cURL command:

curl -fsSL https://raw.githubusercontent.com/lgse/strata/main/install.sh | bash

The installer detects the CPU architecture, the glibc version, and whether the system runs Omarchy 3 or 4 or Arch Linux. It installs the latest stable release for the current user in ~/.local/bin.

Optional extras such as a desktop entry, a default folder handler, SMB support, and Omarchy keybindings are also available. You can even feed Omarchy's onboard AI agent a specific prompt to get it installed on your computer.

If you just want to check out Strata before committing to it, the project's homepage has a demo that lets you play around with it. Building from source is also possible.

Mozilla has a new product called Firefox Enterprise. Which is basically Firefox but built around the needs of private and government organizations.

The idea is simple. If staff uses a browser for work, organizations should be able to control its settings, protect company data, and decide where its management system runs.

Because a lot of work happens directly in the browser. And that's also where the common security issues occur as well. Securing is both a priority and a challenge for organizations. That's where this 'on-prem' version of Firefox comes in.

What is Firefox Enterprise?

Think of it as Firefox with a central control panel for your IT team. Instead of setting up each browser separately, administrators can manage browser rules and access across employee devices.

Firefox has supported business deployments for years, including policy management and Extended Support Release (ESR) versions. This new offering adds a self-hosted management platform, extra security controls, and professional support.

For example, your IT team can decide which extensions employees can use. It can also control browser connections and telemetry, and allow, restrict, or disable AI tools.

There are controls to warn you or block certain actions involving sensitive data, based on the website involved. Security teams can export browser events to their monitoring tools, while a Lockdown Mode adds restrictions for high-risk browsing.

Mozilla says organizations can keep their existing login providers, device management tools, and security monitoring systems. Firefox Enterprise is designed to work alongside them.

The browser still uses Mozilla's Gecko engine. That gives organizations an alternative to browsers built on Google's Chromium technology.

Professional support is included with Firefox Enterprise. If your organization uses regular Firefox, you can buy support separately. Mozilla hasn't listed public prices on the product page, so you'll need to contact its sales team.

Thunderbird Enterprise is coming, too

Mozilla is also working on Thunderbird Enterprise. It plans to let organizations manage email, calendars, and contacts from the same Firefox Enterprise console.

That could make this more useful if an organization already uses both applications. However, Thunderbird Enterprise is still in development, and the announcement doesn't give a release date.

A good fit for Europe's push for digital independence

Mozilla's initial sales focus is France, Germany, Austria, and Switzerland, with more markets planned. Its announcement puts Europe's push for digital sovereignty at the center of the launch.

In simple terms, that's about having more control over your technology and data, rather than relying on a few large suppliers.

I think supporters of that movement will welcome this approach. Running the management console on your own servers gives you more control over where management data and policies live. Using a separate browser engine also helps avoid depending on the same technology everywhere.

Mozilla says the browser is open source, but doesn't confirm whether the management console is open source too. Self-hosted doesn't automatically mean open source.

Another way for Mozilla to earn money

This also gives Mozilla another way to bring in revenue: selling enterprise services and support.

Mozilla still depends on Google through its search partnerships. If Firefox Enterprise attracts paying customers, that income could help reduce Mozilla's dependence on Google.

Of course, it won't replace search revenue overnight, and we don't yet know how much this business focused product will earn.

Still, I find the direction encouraging. Organizations get more control over their browsers, and Mozilla gets a chance to earn more money from the software it already builds. That's a win-win for everyone involved, right?

Remember Ubuntu’s convergence dream? One operating system across your desktop, tablet, and phone, with apps that adapt to whichever screen you’re using.

Canonical pursued that vision with Unity 8 and Ubuntu Phone before pulling its investment in 2017. Now, a new project called Luma wants to tackle a similar ambition.

It’s early days, and only a desktop build is available to download. But its plans go well beyond becoming another desktop Linux distro.

The project's founder, Nick (nmcmil), has worked with a small group of people to achieve this, with the group answering to no shareholders or investors and taking all the key project-related decisions themselves.

When I tested it, it was built on Fedora 44 and ran a recent version of the Linux kernel. The team does say that they have reworked enough of the stack that it doesn't count as a typical distro.

Their blog on AI also points out that the distro features AI-assisted contributions, with contributors being held responsible for any work that carries their name.

Luma: Overview ⭐

The desktop is called Prairie, and the team describes it as a hard fork of GNOME. Its goal is to provide a shell, windows, fonts, and apps that all look and feel the same.

Apps are built with GTK and Libadwaita, with the project shipping LumaUI as a toolkit outside developers can use to make their software feel at home on the system. Luma’s original code is licensed under Apache 2.0, and anything taken from upstream retains its original license.

The distro comes with pre-installed apps built for it; these include Depot as the app store, Viola as the browser, Charlie as the email client, and Tide for any music needs, alongside the self-explanatory Filer, Notes, Photos, and Messages.

If you opt for the Creative Collection, you get access to Canvas for design, Focus for RAW photo editing, Grid for spreadsheets, Stage for slides, Write for documents, Reel for video edits, and Sessions for audio mixing.

You also get Ari, the built-in AI assistant that's meant to work with local models, so everything can stay on your own hardware.

Package support is covered too. I installed apps through DNF5, Flatpak, and Snap on the October 11 nightly, and all three worked.

Valet takes care of AppImage, DEB, EXE, and APK files, which is how Windows and Android apps are meant to sit alongside native ones. It also tells you who made an app and what it can access before you install it.

Android support comes from an optional subsystem based on Waydroid, and compatibility varies by app. Luma is meant to be one system across desktop, tablet, and phone, so apps adjust their layout and controls to the screen size and to touch or mouse input.

There's also the optional Luma Connect app that handles sync for apps, settings, and files across devices.

For Smartphones Too!

Luma also wants to be your smartphone's OS. The team names the Fairphone 6 and the Google Pixel 9 family as its targets and says it plans to support more phones and tablets.

With what I can see from the screenshots they shared with us, the interface looks like it's on par with modern Android and iOS offerings, while sticking to a very measured approach for the colors and elements it employs.

The Luma folks are directing people to their hub's bounty board to vote for the devices they want and pledge money towards getting them supported.

Test it Now

Luma's download page lists a single 7.2 GB x86_64 nightly ISO as the recommended way to try the distro out. It needs a 64-bit Intel or AMD processor with at least two cores running at 2 GHz or faster, 4 GB of RAM, and 40 GB of storage, with an SSD preferred.

An ARM64 build is yet to make an appearance here, so people on that platform will have to wait a bit. Similarly, creative apps like Grid, Stage, Reel, Sessions, and Canvas are also set for a later arrival.

For most of its life, Linux has been built around one simple assumption. There's a box; it has CPUs, memory, and devices, and the kernel manages them. Clusters exist, of course, but they've mostly been a problem for the layers above the OS, i.e., schedulers, orchestrators, distributed frameworks.

AI infrastructure wants more. Today's training and inference setups bind dozens of CPUs, NPUs, GPUs, and memory pools so tightly together that they behave less like a group of servers and more like a single but very large computer. And then come AI agents, which don't just run a model but spin up tools, sandboxes, and long-running sessions on the fly.

That was the core of the keynote Xinwei Hu, Chairman of the openEuler Technical Committee, gave at Open Source Summit Europe 2026 in Prague.

Let's see how openEuler is innovating Linux around agentic scaling.

The problem: the server is no longer the unit

openEuler's keynote slide describes the shift as moving "beyond the box," from a single host to a SuperPoD. A SuperPoD is a tightly interconnected system where compute, memory, and accelerators across many nodes are linked and expected to work as one pool of resources.

That creates trouble the classic OS was never designed for. Memory you need may sit on another node, a fault on one machine can ripple into others, and moving data between hosts and accelerators becomes a bottleneck. And somebody has to manage the whole thing.

Hu grouped openEuler's goals for this into three buckets: ease of use, reliability, and performance.

SuperPoD OS: making a rack look like a server

The ease-of-use goal is ambitious. openEuler wants runtimes and SDKs that work out of the box on SuperPoD hardware, and system administration that is assisted by an agent, so that you can manage a SuperPoD "like a single server."

Reliability is of upmost importance here. According to Hu, openEuler can notify other nodes of a node failure with a latency of 500 ms or less, and can notify individual threads when shared memory develops a fault. It also aims to isolate faults so they don't spread across processes on the same host or across nodes. There's also "fault storm suppression," a fast response to remote memory faults so that a flood of SEA errors doesn't hang the system.

On performance, the Hu's keynote cites two figures. UMDK, which openEuler describes as a unified heterogeneous interconnect infrastructure for SuperPoDs, cuts RPC latency by 20%. An asynchronous pipeline plus unified memory addressing cuts RH2D (Remote Host-to-Device) latency in KV cache operations by 20%.

All of this sits on UnifiedBus, a hardware interconnect protocol in the same family as NVLink, RDMA, and CXL. openEuler's architecture slides show two layers built on top of it. The UB OS Component adds device management, memory pooling, inter-node communication, and virtualization to existing OS components. The UB Service Core adds higher-level services (memory, communication, I/O, a resource engine, virtualization, serverless, and a compiler and runtime).

The part I find most practical is how applications reach this. General-purpose apps keep using the standard POSIX API with transparent calls, while AI applications can use the UB Service Core API directly. If an app needs a memory pool, the UBS Engine finds which servers can spare memory, exports it from them, and imports it into the local machine.

We covered the UnifiedBus foundations in more depth earlier if you want the background.

Where mainline Linux stands

openEuler isn't alone. Mainline Linux has been working on the memory side of the same problem for a few years, mostly through CXL, an open interconnect standard that lets servers attach or share memory beyond what sits on the motherboard.

The idea is to present far memory as just another NUMA (Non-Uniform Memory Access) node, so most applications don't need to change.

That's roughly the philosophy behind openEuler's transparent POSIX path. The difference is scope. Mainline mostly deals with extra memory attached to one host, while SuperPoD OS tries to coordinate memory, devices, and faults across many hosts at once, on a Huawei-originated interconnect rather than a vendor-neutral standard for now.

Agentic Scaling: an OS for workloads that think in loops

The second half of the keynote moves from hardware to how AI itself is scaling. Hu's slide traced three phases. First came "big is better," where intelligence is fixed during pre-training. Then came test-time scaling, where models get smarter through post-training and spending more compute at inference time. The third phase is Agentic Scaling: building new system architecture around the way agents actually behave, while keeping costs under control.

The pitch is aimed squarely at open-weight models (models like DeepSeek, Qwen, and Llama, whose weights you can download and run yourself). openEuler's argument is that this ecosystem won't reach the frontier by chasing raw model size. Instead, it should use the control you get from running models locally to build intelligence at the system level.

Here's what it means. An agent doesn't just call a model once. It runs tools, executes generated code, keeps memory across long sessions, and spawns many short-lived tasks in parallel. Each of those is an OS problem: isolation, startup time, memory management, and recovery.

openEuler's agentic infrastructure slides tackle three of them.

Sandboxes. Agents running code need isolated, disposable environments. openEuler's sandbox stack works with Firecracker, StratoVirt, gVisor, and Cloud Hypervisor, and is compatible with Kubernetes and E2B APIs. Using prewarmed pools and snapshot recovery, the project claims cold starts down to 20 ms (a 70% to 95% reduction), over 2,000 concurrent sandboxes, and task recovery times cut by more than 50%.

KV cache. The KV cache (the model's short-term working memory for a conversation) grows fast with long contexts and many users. openEuler's approach spreads it over four tiers: HBM on the accelerator, local DRAM, a shared DRAM pool across nodes via UnifiedBus, and SSD or object storage for cold data. Combined with a global index that lets one instance reuse cache built by another, the project claims a 30% to 70% drop in time to first token and 1.5x to 3x higher throughput.

Agent memory. For long-term agent memory, openEuler showed oGMemory, built on the openGauss database with vector, full-text, and graph search. The slide claims it halves token overhead and improves task completion by 20%.

What this means for Linux

I think the most important takeaway is that the definition of "the machine" an OS manages is changing, and openEuler is one of the few distros building for that openly.

There's a fair caveat, though. UnifiedBus and the TaiShan and Atlas hardware platforms are closely associated with Huawei's technology ecosystem, and much of the SuperPoD stack is optimized for these platforms.

The real test is how much of this work flows upstream into mainline Linux and plays well with vendor-neutral standards like CXL, rather than staying an openEuler-specific feature. Given that openEuler already counts itself among the top contributors to the Linux kernel, that's the thing worth watching.

Either way, the questions Hu raised (how an OS handles memory it doesn't physically own, faults it didn't cause, and workloads that spawn themselves) are ones every Linux distribution will have to answer soon.

Qiusheng Wu has put out the first release of RustGIS, a new open source GIS application with an announcement on X. v0.1.0 is available right now for anyone to use, along with a second release, v0.2.0, that has rolled out close on the heels of the first one.

The tool's ribbon-based interface will be familiar to you if you have used a modern office suite recently, and the tool itself allows you to analyze map data in various ways.

The release notes for v0.1.0 mention that it comes in 15 interface languages, has support for over 1,000 tools from the GeoLibre and Whitebox suites, and the editing tools cover feature templates and snapping, along with domains and attribute rules.

So, let's check out what RustGIS has to offer. 👇

Who is it for?

GIS stands for geographic information system, which is software for working with data tied to places on Earth. A GIS stacks layers such as roads, rivers, and population on top of each other, matched by location.

Working with those layers in RustGIS starts with analysis tools such as buffer and clip, plus SQL-based selection. Guided workflows cover jobs like viewshed analysis and finding the shortest path. The v0.2.0 release added SQL queries through DuckDB Spatial too.

The app opens common files like GeoJSON, Shapefile, GeoPackage, and GeoParquet, while large files are streamed. You can also open GDAL formats such as File Geodatabase, GPX, and DXF if you install ogr2ogr.

There's AI assistance too! 🤖

The AI Assistant is a chat pane that answers questions about the data or runs commands on the map. Undoing a change takes one step, and commands that touch files or the network wait for approval.

The command line can also render maps and chain tools, while an MCP server lets agents such as Claude Code and Cursor run RustGIS commands and generate maps.

Download RustGIS

If you are looking to try it out, RustGIS is suitable for terrain and hydrology work and for anyone who wants to execute repeatable workflows from the command line. Newcomers might also find the assistant's plain-language prompts easier than wading through the menus.

The newest release at the time of writing was v0.2.0. You can visit the downloads portal to grab binaries for Linux, Windows, and macOS. If you prefer a web-based solution, RustGIS is also offered as a WebApp.

The project hosts the source code for this on GitHub, and it also has a comparison table that shows how RustGIS fares compared to other tools like ArcGIS Pro and QGIS.