Red Hat’s latest release, RHEL 10.3, puts AI front and center. Its downstream relatives have different priorities.

A week ago, Red Hat released Red Hat Enterprise Linux 10.3, and the clones got busy. Five days later (or three “working” days), the AlmaLinux Foundation released AlmaLinux 10.3 and CIQ released Rocky Linux 10.3. None of this was much of a surprise.
Late last week, when I started working on an article about RHEL’s release I reached out to each of the other two projects to see when they would upgrade their distros to be reliable copies of RHEL’s latest. benny Vasquez, chairperson at AlmaLinux, told me they were anticipating a release on Monday, which happened right on time. That also wasn’t a surprise. For a while now, AlmaLinux has been releasing its clone within a day or three of Red Hat, and a beta of AlmaLinux 10.3 had been out for longer than a month.
Rocky Linux’s answer was a bit more of a surprise, since that distro has traditionally come a week or more after a new RHEL release, although lately the distro has been tightening the gap. Lindsay Aamondt, CIQ’s VP of Marketing, told me that Rocky Linux was shooting for a release on Wednesday, and anticipated my next question with, “thanks to improvements in our testing process including Testarc and efforts by Testing Lead Chris Stackpole (Stack) to improve test automation with Sparrow6 and Sparky. As always, this date is a target and not a guarantee.”
As it turned out, there was no need for the closing caveat. The release came two days early, on the same day as AlmaLinux. Hooray for Rocky.
With the two major clones falling in line so quickly with releases that are basically different brands of the same software, I think it’s time to quit covering their releases as separate events, but cover Enterprise Linux releases as a single event, focusing on Red Hat’s initial release and then making note of any differences the brands bring to the table. We’re also not going to pay much attention to Wednesday’s release of RHEL 9.9, which basically gives the RHEL 9.x series some parity with the latest and greatest, nor with Monday’s subsequent 9.9 releases from the clones.
I’m pretty sure that’s an approach the open source clones will appreciate, and that IBM… er, Red Hat, will despise.
Red Hat’s Enterprise 10.3 in a Nutshell
Software vendors are always prognosticating about the future on release day, so much so that release announcements usually carry a legal reminder to not take “forward looking statements” too seriously, because they’re basically just educated guessing. For Red Hat, that seems to be especially true with this release, as its press release doesn’t deal much with the old familiar traditional OS design ingredients that once dominated a Red Hat release announcement. Instead, with this release it seems to be selling its expertise to turn what many still regard as tomorrow’s technologies into practical tools for today.
The keyword here is AI, of course. Like everybody else in tech these days, Red Hat leads with AI, which it’s both harnessing as a tool, and defending against. For the latter, it advocates fighting fire with fire — and starts by playing the scare card.
“AI-driven threats move with unprecedented speed, which means traditional security is no longer enough,” Gil Cattelain, Red Hat’s principal product marketing manager, said in the release’s availability announcement. “Red Hat has shifted toward a model of proactive resilience, where security is fundamentally integrated with intelligence.”
In other words, Red Hat’s fighting AI with AI.
“By using AI-augmented diagnostics to predict vulnerabilities and implement automated, intelligent remediation, we are defending infrastructure while actively hardening it against the next generation of sophisticated threats,” he added.
There’s plenty to talk about in this new release:
- Sealed images for image mode, introduced as a developer preview in 10.2, are now generally available. These add cryptographic integrity verification from boot through runtime. Teams can build, sign, and seal an approved RHEL configuration, then deploy that image across a server fleet, or build different images for different server roles. Sealing adds assurance that the protected operating-system content matches what was approved, with tampered content rejected when it’s read.
- Model Context Protocol server, another dev preview that’s now GA. This one wrangles AI assistants, or at least connects them directly to the distro’s system data to allow “intelligent” (because AI) automated diagnostics. This one doesn’t actually show up ready to run, but as an RPM in the RHEL extensions repository, allowing you to install it alongside goose, Red Hat’s AI command-line agent.
“By integrating historical Performance Co-Pilot (PCP) data, your AI assistants now analyze past performance events to more quickly identify root causes of intermittent issues, without the need for manual, tedious data correlation,” Cattelain said.
- Increased self-managed command-line assistant capabilities introduced as a developer preview. While RHEL’s command-line assistant remains the natural-language assistant for the terminal, users can now operate the assistant in a self-managed way, using their own hosted inference, helping keep model guidance, sensitive operational context, and digital sovereignty secure. Red Hat’s hosted path remains available for connected estates.
- Developer enhancements to accelerate modern application development. The new release ships with updated developer toolsets, including GCC 16, LLVM 22, and Go 1.27. Also included are .NET 11, Node.js 26 (LTS), and PHP 8.4 (the latter on 9.9 only, evidently). For the security minded, Red Hat cites Rust 1.97, and new grafana-pcp AI dashboards to unify GPU, CPU, and network performance telemetry, and simplify monitoring for AI/ML teams.
“AI has fundamentally changed enterprise security, reinforcing that critical workloads are only as resilient as the underlying foundation they run on,” Raj Das, GM of RHEL, said in a statement. “Red Hat Enterprise Linux delivers this reality, anchoring zero trust principles and threat resilience at the heart of modern workloads.”
So, the word of the day as far as the RHEL folks are concerned is AI.
The Release of the Clones
Almost everything about Enterprise 10.3 that’s getting a big sales pitch from Red Hat is absent from both Rocky Linux’s and AlmaLinux’s press releases. Not only that, I’m unable to verify that much of what’s contained in the bullet list of new RHEL features is even included in the cloned versions.
It appears that neither clone includes MCP server or sealed images, although Rocky Linux does support building bootc-compatible, image-mode-style assets. The self-managed command-line assistant that Red Hat includes in developer mode also isn’t announced as a Rocky or AlmaLinux feature.
The items listed with the fourth bullet seem to be largely included on Rocky and Alma, but with some qualifications around Red Hat’s grafana-pcp AI dashboards. Rocky lists prebuilt dashboards for NVIDIA GPU metrics and vLLM, which doesn’t completely match Red Hat’s wording, so it’s the same or something close. AlmaLinux 10.3 includes Grafana 12.4.3 and PCP 7.1.5, but its release notes don’t specifically list Red Hat’s new Grafana PCP AI dashboards for GPU, CPU, and network telemetry.
I’ve reached out to Rocky Linux and AlmaLinux for clarification, but at press time haven’t heard from either. Until I do, the only conclusion I can draw is that what Red Hat wants you to think is important about Enterprise 10.3 and what Rocky and AlmaLinux think is important about the release aren’t the same. If that assessment is correct, I tend to side with the clones. A secure, rock-solid Linux distribution is more important than peripherals that are bolted on.
Christine Hall has been a journalist since 1971. In 2001, she began writing a weekly consumer computer column and started covering Linux and FOSS in 2002 after making the switch to GNU/Linux. Follow her on Twitter: @BrideOfLinux




Be First to Comment