Models & Privacy
The workspace stays. The model is your choice.
NexxusOS is not an AI model. It is the workspace around supported models — with its own records, tools and controls. The model you choose still affects answer quality and capability.
Two kinds of model route
Your NexxusOS workspace
- Conversations & memory
- Files & projects
- Tasks & approvals
- Settings
Stays in place when the model changes.
Supported local models
Run on your Windows device.
Configured providers
Connect a provider you choose.
Supported local models
Run on your Windows device. Requests handled by a local model are processed on your hardware. What a local model can do depends on the model and on your device.
Configured providers
Connect a provider you choose. Provider usage is billed by that provider under their terms, and their own policies apply to what they receive.
What is sent, and when
When you allow a request to use a configured provider, the content needed for that request is sent to that provider. Local-first does not mean cloud requests carry no content.
Web research and other connected services also need internet access and send what they need to work. Using a local model for a request does not, by itself, make every other connected feature offline.
Your workspace — files, projects and approved memory — is stored on your Windows device. Local storage is a statement about where your workspace lives; it is not a claim of encryption or anonymity.
Continuity without overpromising
Your conversation records, projects and approved memory belong to the NexxusOS workspace, so switching model doesn't mean starting again. Different models still have different abilities and context limits, and the models themselves don't learn about you from your workspace.
For the full notice covering this website and the desktop app, read the privacy notice.