Run Duvo’s AI on EU-only, single-tenant GPUs, with other geographies available by agreement. You can also use your own infrastructure or an approved AI endpoint.
Run open models on dedicated GPUs reserved for your organisation, with inference kept in the EU. Dedicated inference is also available in other agreed geographies.
Duvo arranges the capacity. We agree the model, operating responsibilities and support with your team.
Run a compatible model in your own cloud or on-premises environment, and connect Duvo to it.
Your team operates the infrastructure. We validate the model against your process and agree connectivity, access and support.
Use a compatible AI endpoint your company already approves, including a managed model service.
Use the provider and data-handling terms your team has approved. We confirm model compatibility and the operations it will support.
The proposal separates GPU capacity, setup and operations from the process service. We evaluate model quality on your process before agreeing the deployment.
GPU location defines where model inference runs. The deployment proposal also specifies where orchestration, system actions, files and records live, and who can access them.
On-premises inference runs the model on your infrastructure; it does not move the whole Duvo platform there. Your security team reviews the processing locations and provider terms before deployment.
Duvo does not train its own or third-party models on customer data.
Review security and data handling