The latest release enhances a local artificial intelligence server with speech synthesis, three-dimensional generation, improved Linux neural processing unit support, and expanded model registry support.
AMD announced the launch of Lemonade 11.0, the latest version of its open-source local AI server for running large language models and local AI workloads on AMD Ryzen processors, Radeon graphics processors and Ryzen AI neural processing units (NPUs) on Windows and Linux. The upgrade brings in some exciting new features that enhance local AI development while making deployment easier.
One of the prominent new features is text-to-speech (TTS), which is now available as an OpenMOSS backend option. TTS comes directly into the Lemonade interface and supports voice cloning and voice design models, allowing both to run locally without relying on cloud services.
Another exciting new feature added to Lemonade is 3D content generation via integration of Trellis.2 image-to-3D generation pipeline. Users can now generate 3D content using API endpoint and preview it in the integrated model viewer.
With the release of Lemonade 11.0, ModelScope has been added as a second remote model registry alongside Hugging Face. Users can choose between model registers using the desktop application, CLI, or API.
For Linux, the new release enables better deployment because it automatically installs the FastFlowLM NPU backend by default, which removes the need for users to install system packages manually. It also enables improved request routing since the collection router model will choose the right models for inference reports the selected inference route during execution.
As an open source platform, Lemonade remains relevant not only as a demonstration platform for local AI acceleration but also as an environment to run modern AI models locally. With the support for CPUs, GPUs, and NPUs along with speech, 3D generation, and deployment improvements, Lemonade 11.0 provides additional functionality to developers who build local AI applications.
