Base Compute

Base Compute builds runtimes and infrastructure for on-device AI inference. The team operates from Melbourne and Berlin, working at the runtime and silicon level to make powerful AI run locally on hardware people already own. Their flagship product, BaseRT, delivers the fastest LLM inference on Apple Silicon through hand-written Metal kernels and automated research pipelines. Base Compute also targets enterprise deployments with on-premise, hybrid, and air-gapped AI infrastructure.