The open-source Lemonade Local AI Server, which enables LLM with Ryzen AI NPUs on Linux, enables AMD Radeon GPU support and popular x86_64 CPU support (in addition to Microsoft Windows support), making it easy to embed into other apps to consume AI.
Lemonade continues to advance rapidly with local AI server capabilities that primarily support AMD CPUs/GPUs/NPUs, and the company’s engineers are heavily involved in its development. Just a few days after the release of Lemonade 10.1, Lemonade 10.2 was released with a focus on making it more embed-friendly.
Lemonade 10.2 exposes embeddable Lemonade release artifacts. These builds for Linux and Windows only include the Lemond daemon, Lemonade CLI, and related resource files, but do not include web apps, Electron bits, or other code not required for embedding use cases. We also have new documentation outlining embeddable usage of Lemonade with runtime integration, backend/model support, and more.
“Embeddable Lemonade is a binary version of Lemonade that you can bundle with your own apps to provide a portable, auto-optimized, multimodal local AI stack. This allows users to focus on their apps without the need for the Lemonade installer, branding, or telemetry.”
This embeddable Lemonade continues to support LLM across GPU, NPU, and CPU. Learn more about the new embeddable features via a pull request from AMD’s Jeremy Fowers, who introduced new embeddable release artifacts. There’s also new documentation explaining the use of embeds. Remember that Lemonade is open source under the Apache 2.0 license.
Lemonade 10.2 also includes improved integrations, including automatic download of GGUF and RAI models, support for Qwen image models, and OpenCode integration.
Downloads and details for Lemonade 10.2 are available on GitHub.
