ARK Labs

ARK Labs provides inference infrastructure designed for EU data residency and AI Act compliance. The platform supports heterogeneous GPU fleets (NVIDIA RTX 4090/A100/H100, AMD MI300X/MI250X, Intel Gaudi 3) and pools different GPU generations without requiring specialized networking like NVLink or InfiniBand. GPUs can be added or removed live without service interruption.

Three deployment models are available: ARK Cloud (managed EU-hosted API with pay-per-token pricing), ARK Core (self-hosted on your own hardware), and ARK Tailored (self-hosted with add-ons for vision, speech, and embeddings). The API is OpenAI v1 and Anthropic compatible. Includes stateful inference for agent workflows and audit-ready logging.