Skip to main content

Inference service connection types

Last update:

When creating an inference service, you can choose one of the following connection types: public access or private access.

Public access

When public access is selected, a free domain name is automatically generated. You can connect to the inference service from the internet via a public endpoint.

Private access

When private access is selected, a new private subnet is automatically created, where the inference service is placed. You can connect to the inference service via a private endpoint only from devices located in the same private subnet. For example, you can configure a private connection to the inference service via a cloud server.

In the private subnet of the inference service, three IP addresses are reserved for internal service needs and cannot be assigned to other devices. These are typically the first IP addresses in the subnet range, for example: 172.16.0.1, 172.16.0.2, 172.16.0.3.

You cannot connect an inference service with private access to devices in other networks and pools via a cloud or global router.