WasmEdge

Why we included this project

Running large language models usually means renting a cloud GPU, but WasmEdge is aimed at teams that want those workloads closer to the data: on personal machines, edge nodes, or constrained IoT hardware. It is a WebAssembly runtime that executes GenAI models, including LLMs, speech-to-text, and image generation, on both CPUs and GPUs across servers, laptops, and edge devices, and it is the engine behind the LlamaEdge framework. Because models run inside a single runtime instead of depending on host libraries, the same pipeline can move from a developer's laptop to a cluster to an embedded target without the usual environment breakage. It also works as a general sandbox for safely executing untrusted code, which is useful if you want to host third-party or user-contributed model wrappers. Before committing, check the AI extensions and LlamaEdge documentation to confirm your model set and hardware targets are covered.