What you can build

cLLMHub is an OpenAI-compatible API for open models. Use it as a hosted backend, or publish your own — same endpoint either way.

Serve your home GPU as an API

Got a workstation or gaming PC with a decent GPU sitting at home? Run cllmhub publish to expose its Ollama, vLLM, llama.cpp, or MLX backend through a stable, public, OpenAI-compatible URL. No port forwarding, no dynamic DNS, no reverse proxy.

Call your fine-tuned model from anywhere

Trained a specialist model — code, legal, medical, a specific language? Register it once with the CLI and call it over the standard OpenAI API from any app or environment.

Drop-in OpenAI replacement for your own model

Point any OpenAI SDK at api.cllmhub.com/v1 with the model name you published. Same request shape, same streaming, same tool calls — your code keeps working, but the model is yours.

Apple Silicon as a personal inference server

Run MLX or Ollama on an M-series Mac and use it as your private inference backend from anywhere — laptop, phone, work machine — all through the same hosted API.

Private RAG over your own data

Embeddings, summarization, and chat all running on a model you control. Your documents never touch a third-party inference provider.

Background workflows on your hardware

Cron tasks, queue workers, and data pipelines that hit your own LLM. Flat monthly cost — high-volume batch work doesn't blow up the bill.