Earlier this month with the release of the Lemonade SDK 10.0 and FastFlowLM 0.9.35, using AMD Ryzen AI NPUs for running LLMs on Linux finally became feasible. AMD XDNA 2 NPUs can now run on Linux well for LLM workloads! Released on Tuesday was Lemonade 10.0.1 with a few improvements for the setup process of this local LLM open-source solution on Linux.
The Lemonade SDK helps Windows, macOS, and Linux users run large language models on their GPUs, CPUs, and NPUs. With Lemonade 10.0.1 the Debian packages are now available via a Personal Package Archive (PPA) for easier installation on Ubuntu Linux. There is also a smoother FastFlowLM install process for Linux users.
There are also updates to the Linux NPU instructions, a FastFlowLM setup guide for Arch Linux use, Lemonade now supports system tray support using AppIndicator3, and Fedora install documentation too.
Beyond the Linux setup/install improvements, Lemonade 10.0.1 streamlines the process of searching and adding GGUFs from Hugging Face, Qwen3.5-4B support on NPUs using the latest FastFlowLM, and updating the bundled Llama.cpp version.
Downloads and more details on the Lemonade 10.0.1 release via GitHub.
