Browse Endpoints

View the list of vLLM endpoints. The screenshots use the CPU menu; the GPU and Rebellions menus are the same except for the resource column (GPU / NPU).

Prerequisites

• An account with access to AI Inference
• The cluster and project to view, selected at the top of the sidebar

---

Sidebar

Open the endpoint list for each compute type (GPU, CPU, Rebellions) from the vLLM group. The three screens are the same except for the resource column.

No.	Feature	Item	Description
1-1	GPU vLLM	GPU menu	· Opens the screen for vLLM endpoints running on GPU.
1-2	CPU vLLM	CPU menu	· Opens the screen for vLLM endpoints running on CPU.
1-3	Rebellions vLLM	Rebellions menu	· Opens the screen for vLLM endpoints running on Rebellions NPU.

Search · Bulk delete

Find endpoints by name and delete selected endpoints at once.

No.	Feature	Item	Description
2-1	Search	Search box	· Filters the list by endpoint name.
2-2	Bulk delete	Delete button	· Deletes the endpoints checked in the table. Enabled only when at least one row is selected.

Create endpoint

Opens the screen for creating a new endpoint.

No.	Feature	Item	Description
3-1	Create endpoint	Create endpoint button	· Click the button at the top right to open the endpoint creation screen.

Endpoint list

Shows the endpoints of the current project and cluster in a table.

No.	Feature	Item	Description
4-1	Columns	Status	Endpoint status. Green means running, gray means paused (including scaled to zero with no requests), and red means an error.
		Name	Endpoint name, with the deployed model (or image) below it.
		CPU	CPU resources allocated to each replica.
		Replicas	Minimum - maximum replicas. With a minimum of 0, the endpoint scales down to zero when idle.
		Port	Port on which the endpoint receives requests.
		Created at	Date the endpoint was created.
4-2	Row actions	Action (⋯) button	· Click the button at the end of a row to open that endpoint’s action menu.

Row action menu

Actions available for each endpoint. Some items are disabled depending on the endpoint status.

No.	Feature	Item	Description
5-1	Menu items	Service URL	View and copy the endpoint URL.
		View logs	View pod events and container logs.
		Send prompt	Send a test prompt to a running endpoint and check the response.
		Pause	Pause a running endpoint.
		Resume	Resume a paused endpoint.
		Cancel	Cancel an endpoint deployment that is in progress.
		Edit	Open the screen for editing replicas and resources.
		Delete	Delete the endpoint. A confirmation modal opens first.

Endpoints with a minimum of 0 replicas

An endpoint with Replicas 0 - N scales down to zero when idle and may appear gray (paused). It starts again automatically when a request arrives.