Configure public connection to an inference service via Open WebUI Virtual Machine

You can configure a public connection to the inference service via Open WebUI Virtual Machine. This will allow you to interact with the model in the Open WebUI web interface using a public endpoint via the internet.

For configuration, an inference service in public access is used. Open WebUI Virtual Machine must be created with access from the internet:

in a public subnetwork;
or in a private subnetwork with a single public IP address.

Create a cloud server with Open WebUI.
Create an administrator account in Open WebUI.
Run Open WebUI.
Connect an endpoint in Open WebUI.
Send a test request to the model.

1. Create a cloud server with Open WebUI

In the control panel, on the top menu, click Products and select AI Marketplace.
Click Create server.
Fill in the blocks:
- Name and placement;
- Source;
- optional: GPU;
- Configuration;
- Disks;
- Network;
- optional: Access;
- optional: Additional settings;
- optional: Automation.
Check the server price.
Click Create server.

Name and placement

Enter a server name.
Select a location where the server will be created. The list of available GPUs depends on the location. You cannot change the location after the server is created.

Source

Select the Open WebUI VM (Ubuntu 24.04 LTS 64-bit) image.

Optional: GPU

Click Add GPU.
Select the GPU type. When choosing a GPU, consider the minimum resource requirements for the image to work. GPU specifications and descriptions can be found in the Graphics Processing Units (GPU).
Specify the number of GPUs.

After the server is created, you will be able to change the GPU type and quantity or delete the GPU. For details, see the Changing Cloud Server Configuration.

Configuration

Specify the number of vCPUs.
Specify the RAM size.

After the server is created, you will be able to change the configuration.

Disks

Select the boot disk type. GPUs are not available when using a local disk as a boot disk.
Specify the disk size in GB or TB. The maximum size for all network volumes is 10 240 GB (10 TB), for a local disk — 1 256 GB (1 TB).
If you selected SSD Universal v2 or SSD Fast v2 disk type, specify the total read and write operations in IOPS. After the disk is created, you can change the number of IOPS — decrease or increase it. The number of IOPS changes is unlimited.
Optional: to add additional disks:

4.1. Click Add disk.

4.2. Select the disk type.

4.3. Specify the disk size in GB or TB. The maximum size for all network volumes is 10 240 GB (10 TB), for a local disk — 1 256 GB (1 TB).

4.4. If you selected the SSD Universal v2 or SSD Fast v2 disk type, specify the total read and write operations in IOPS. After the disk is created, you can change the number of IOPS — decrease or increase it. The number of IOPS changes is unlimited.

After the server is created, you will be able to detach additional disks or attach new ones.

Network

To set up public connectivity for the inference service, the server must be added to a subnet with internet access:

a private subnet with a public IP address. A public floating IP address will be connected to the server's private IP address, and internet traffic will be routed via NAT 1:1 through a cloud router;
or a public subnet where all addresses are accessible from the internet.

You can select an existing subnet or create a new one.

Private subnet with a public IP address
Public subnet

If you already have a private subnet and a cloud router with internet access:

1.1. In the Subnet field, select the subnet type Private + 1 public IP.

1.2. In the Private field, select an existing subnet.

1.3. Optional: change the server's default private IP address.
If you want to create a new subnet and a cloud router with internet access, in the Subnet field, select the subnet type Private + 1 public IP.

A private network nat, a router router-nat, and one public address will be created automatically. The server will be added to the subnet in the nat.

Optional: Access

In the Password for root field:

1.1. Copy the root user password — the user with unrestricted privileges for all system actions.

1.2. Save the password in a secure place and do not share it in plain text.
Place an SSH key for the project on the server for secure connection:

2.1. If the SSH key is not added to the cloud platform, click , enter the key name, paste the public key in OpenSSH format, and click Add.

2.2. If the SSH key is added to the cloud platform, in the SSH-key field, select an existing key.

Optional: Additional settings

To create a preemptible server, check the Preemptible server.
If you plan to create several servers and want to increase infrastructure fault tolerance, add the server to a placement group:

2.1. To create a new group, click , enter the group name, and select a placement policy on different hosts:
- preferred — the system will try to place servers on different hosts. If no suitable host is available when creating a server, it will be created on the same host;
- required — servers in the group must be located on different hosts. If no suitable host is available when creating a server, the server will not be created.
2.2. If the group is already created, in the Placement group field, select a placement group.
To add additional information or filter servers in the list, add server tags. A tag with the image name is added automatically. To add a new tag, enter it in the Tags field.

Optional: Automation

To add a script that will be executed by the cloud-init agent during the first operating system boot, in the User data:
- open the Text tab and paste the script as text;
- or open the File tab and upload the file containing the script.
Examples of scripts and supported formats can be found in the User data.

2. Create an administrator account

To work in the Open WebUI web interface, first create an administrator account.

In the control panel, on the top menu, click Products and select AI Marketplace.
In the Open WebUI Virtual Machine card, click Go to GUI.
In the Open WebUI web interface, click Let's start.
Enter a username.
Enter an email address.
Enter a password.
Click Create Administrator account.

3. Launch Open WebUI

In the control panel, on the top menu, click Products and select AI Marketplace.
In the Open WebUI Virtual Machine card, click Go to GUI.
Enter the administrator email address.
Enter the administrator password.
Click Sign in.

4. Connect an endpoint in Open WebUI

In the Open WebUI web interface, click the profile icon and select Admin Panel.
Open the Settings.
Go to the Connections.
Turn on the Direct connections.
Click Save.
On the Open WebUI web interface home page, click the profile icon and select Settings.
Go to the Connections.
In the Manage direct connections row, click .
In the URL field, enter the endpoint:
```
<endpoint>/v1
```
Specify <endpoint> — the inference service endpoint; you can copy it in the control panel: on the top menu, click Products → Inference Services → in the inference service card, in the Inference service endpoint row, click .
In the Bearer field, enter the API key. You can copy the API key in the control panel: on the top menu, click Products → Inference Services → inference service page → API Keys tab → in the API key row, click , and then .
In the Model IDs field, enter the model name and click . The model name can be viewed in the control panel: in the top menu, click Products → Inference Services → inference service page → tab Service → line Model.
Click Save.

5. Send a test request to the model

In the Open WebUI web interface, click Select a model.
Select the model you added in step 8 when connecting the endpoint in Open WebUI.
On the Open WebUI web interface home page, click the profile icon and select Settings.
Go to the General.
In the Advanced parameters block, set the max_tokens parameter value to 5 000.
Click Save.
Enter a test request, for example:
```
Hello!
```
Click .
Wait for the model response.