Your own LLM cloud,
on the machines you already have.

Install a small agent on your servers and workstations. They join one fleet behind an OpenAI- and Anthropic-compatible API, and you decide which open models run where.

Sign in

One command per machine

The agent dials out to the control plane: no open ports, works behind NAT. It picks the right engine build for CUDA, Apple silicon or plain CPU.

Models where they fit

Browse Hugging Face, add a model, deploy it to the nodes with room for it. Downloads resume and are verified; progress shows live.

Drop-in API

OpenAI (Chat, Responses with server-side conversations, Embeddings), Rerank, Anthropic Messages and Ollama. Streaming and tool calls included.

Smart routing

Requests go to a node that already has the model loaded, by priority tier, and never to a workstation while its owner is using it.

POST/v1/chat/completionsOpenAI Chat Completions
POST/v1/responsesOpenAI Responses
POST/v1/messagesAnthropic Messages
POST/v1/embeddingsOpenAI Embeddings
POST/v1/rerankRerank documents for search (Cohere and Jina format)
POST/api/chatOllama API, for apps that only speak Ollama
GET/v1/modelsModels served right now