Linux users who've been waiting for a mainstream AI desktop app to rival the likes of Alpaca and Moose can finally breathe a sigh of relief - just as long as they don't mind keeping their AI in the cloud. Anthropic's new Claude Code desktop app for Linux brings all the features of its MacOS and Windows siblings to the penguin-powered masses, but with a catch: it really doesn't want to play nice with your local AI setup.

ZDNet's resident Linux enthusiast gave the app a thorough test drive, installing it via the required Debian/Ubuntu repository (because nothing says "welcome to Linux" like a terminal command). After adding Anthropic's signing key and repository, the app installed cleanly and presented a well-designed GUI - exactly what you'd expect from a polished commercial product. But the trouble started when our tester tried to connect it to a locally installed Ollama instance, which is how many Linux users prefer to roll with AI. Despite enabling Developer options and configuring third-party inference, the app stubbornly refused to see any locally pulled models, leaving our tester stuck with a free Anthropic plan that's about as generous as a vending machine that only accepts exact change.

On the bright side, the Linux app is functionally identical to the MacOS version - a rare feat of cross-platform parity that would make even a seasoned QA engineer nod in approval. And when pitted against a local AI app running the same query ("Write a Linux GUI app for Pop!_OS COSMIC Desktop for creating invoices..."), the cloud-dependent Claude Code kept the system humming along smoothly, while the local alternative brought the machine to a grinding halt. So if you've got a paid Anthropic account and a hankering for simplicity, Claude Code Desktop is a solid choice. But if you value privacy and local control, stick with Alpaca or Moose - they may not have the brand recognition, but at least they don't require you to sell your data to the cloud gods.