Local model
A local AI model is a model that runs on your own hardware rather than through a provider’s cloud service. Prompts and answers stay on your machine, and no outside service sees them.
In a run
You set which model handles which data. A step that reads a drawing or a client file can go to a local model while other steps use another. Each call goes to the model you chose, and the record shows where each call ran.
Our word for it
Runs in your building.
Your choice of model
Ares does not tie your agents to one model. They use the models you choose, local models included. A whole Ares site, with its local models, runs on one small machine in your building.
A local model is often the reason a workflow can go ahead at all. When the data may not leave the building, the model has to come to the data.
You can mix models in one workflow. The steps that touch controlled data can use a local model, and the other steps can use another. A local model also suits the disconnected shape, where no step can call out to a cloud service.
Where it stops
Local models differ from the largest cloud models. A step that needs the largest model may not suit a local one. We say which model runs which step, and where, so the choice is yours and on paper.
Ares also does not make a model more accurate. It sets what the agent may do with the answer, and it records the decision.
See the deployment page, or read the five guarantees and where each one stops.
Related terms
- On-prem AIOn-premises AI is AI that runs on hardware inside your own building instead of in a vendor’s cloud.
- Data residencyData residency is the rule that data must be stored and processed inside a set country or region.
Tell us about one workflow.
We will say whether an agent should do it, and what the first step costs.
A senior member of the team replies within two business days.