AI Inference | Private LLM

Language models,
inside your boundary.

Run open models in your own datacenter or on the device itself. Prompts and outputs stay inside your network, and Supercloud keeps the models served and up to date.

Talk to our teamCompare with Token Factory

Where it runs

Your datacenter, or the device itself

Your data

Prompts and outputs stay inside your network

Models

Chosen with your team

Everything stays inside your boundary

Your applications call an endpoint on your own network. Prompts, documents and answers stay on machines you control.

your networkanswers stay on your networkYour appPOST /v1/chat/completions{ "model": "your-model", "messages": [ … ]}Private endpointroutes by modelyour datacenterChat modelOpen weightschatEmbedding modelOpen weightsembedon the deviceOn-device modelSmall open modeledgeyour networkYour appPOST /v1/chat/…"model": "your-model""messages": [ … ]Private endpointroutes by modelyour datacenterChat modelOpen weightsEmbedding modelOpen weightson the deviceOn-device modelSmall open model

What you get

  • Your data stays yours

    Prompts, documents and outputs stay on your machines and aren’t used to train models.

  • From the datacenter to the device

    Serve larger models on your own servers and smaller ones on devices that have to work offline.

  • Run like a managed service

    Supercloud deploys, updates and watches the models, so your team doesn’t run inference servers by hand.

The same call, on your own network

Applications send requests in the chat-completions format most SDKs already speak.

  • Open models you choose

    Pick open chat, code and embedding models, or bring a fine-tune of your own.

  • Keys and roles

    Project roles decide who can deploy a model and which keys can call it.

  • Usage in the console

    Follow requests and token usage per application and per model.

POST /v1/chat/completionscurl
curl https://llm.your-company.internal/v1/chat/completions \
  -H "Authorization: Bearer $PRIVATE_LLM_KEY" \
  -d '{
    "model": "your-model",
    "messages": [
      {"role": "user", "content": "Summarize this contract"}
    ]
  }'
Served from machines on your network.

FAQ

How is it different from Token Factory?

Token Factory serves open models from our cloud through one API. Private LLM serves them on your own infrastructure or on the device, so requests stay on your network.

Which hardware does it run on?

Tell us what you have and we’ll confirm what fits.

Can it run without internet access?

Tell us how your network is set up and we’ll walk you through it.