Fedora or Omarchy for a Local AI Workstation

Fedora and Omarchy are both good foundations for local AI. Hardware and the inference runtime determine model performance; the operating system shapes how the machine feels to run and maintain.

Fedora is the calmer infrastructure choice. Omarchy is the more opinionated desktop choice.

Why Fedora works

Fedora gives me recent kernels, Mesa, Python, Podman, and strong security defaults without becoming a rolling release.

That matters on new AMD hardware. Vulkan, memory management, and amdgpu fixes often arrive through the kernel and Mesa.

Fedora also keeps the base system fairly predictable. SELinux is enabled, services use systemd, and container tooling is already part of the platform.

For a GTR9 Pro that spends most of its life serving models over Tailscale, that is exactly what I want.

Install the runtime, configure the service, and leave the machine alone.

Why Omarchy works

Omarchy is an opinionated Arch-based desktop built around Hyprland. It arrives with a complete developer environment ready to use.

Its menu can install Ollama or LM Studio, and it includes coding-agent workflows through tools such as OpenCode. Docker, editors, terminals, and development environments are already integrated.

Arch also gives Omarchy current kernels, drivers, and access to the AUR. That is useful when support for new hardware or a new inference tool has not reached slower distributions.

The trade-off is the rolling base. Updates arrive quickly, but they require more attention.

If the AI machine is also my daily desktop, Omarchy is more interesting than a plain workstation installation.

GPU support still needs work

The GPU stack remains the hard part on either distribution.

ROCm, Vulkan, NVIDIA drivers, and model runtimes still depend on exact hardware and software versions. A newer package can fix one workload and break another.

I start with llama.cpp or Ollama, confirm one model works correctly, and only then optimize the GPU stack.

My choice

For a dedicated inference server, I choose Fedora.

For a workstation I will use directly every day, I would happily choose Omarchy.

Both are current enough for local AI. The right choice depends on whether I want the operating system to stay in the background or become part of the experience.