AMD Lemonade SDK 10.7 Adds NVIDIA CUDA Support — Open-Source Local AI Server Now Runs Across AMD, NVIDIA, and Apple GPUs

AMD released Lemonade SDK 10.7, adding NVIDIA CUDA support via Llama.cpp's CUDA backend on Windows and Linux — making this open-source local AI server the first AMD-developed tool to run natively across AMD, NVIDIA, and Apple Metal GPUs. Version 10.7 also adds stable-diffusion.cpp CUDA and Vulkan backends for broader cross-vendor image generation, LMX-Omni model support, a Prometheus metrics endpoint for real-time monitoring, and a 'lemonade bench' command for standardized LLM benchmarking across Llama.cpp, vLLM, FastFlowLM, and Ryzen AI. The SDK is OpenAI, Anthropic, and Ollama API-compatible.