Lumabri

Lumabri is a private inference service that runs large AI models distributed across multiple computers on a local network. It manages memory allocations, model weight distribution, and execution layers to enable running supported models larger than a single machine's capacity. It provides chat interfaces via terminal and bundled web page, supports explicit resource sharing approvals, and includes tools for performance measurement and debugging. Lumabri is built in C and integrates with Colibri for model execution.