Turn NVIDIA GPUs you own into a private, OpenAI-compatible LLM API — for a homelab box or a team's rack server. Mainline vLLM and llama.cpp behind one port, a key per app, honest per-GPU monitoring, and Claude Code and Codex CLI routed to models on your own cards. Formerly LLM Warden.
About
LM Warden turns NVIDIA GPUs you own into a private LLM API that works like a hosted one: a base URL, a key, and the OpenAI client library you already use. It runs the two most widely used open inference engines, vLLM and llama.cpp, unmodified, behind one published port and a brow…
