Contact UsGet started
© 2025 Direct Cursus Technology L.L.C.

Creating a VM with a GPU

Written by
Improved by
Updated at May 5, 2025

This section explains how to create a VM with a GPU. For more information about VM configurations, see Graphics processing units (GPUs).

By default, a cloud has a zero quota for creating VMs with GPUs. To change the quota, contact support.

You can create VMs based on Intel Broadwell with NVIDIA® Tesla® V100, Intel Cascade Lake with NVIDIA® Tesla® V100, and AMD EPYC™ with NVIDIA® Ampere® A100 in the ru-central1-a and ru-central1-b availability zones.

  1. In the management console, select the folder where you want to create your VM.

  2. In the list of services, select Compute Cloud.

  3. In the left-hand panel, select Virtual machines.

  4. Click Create virtual machine.

  5. Select a GPU-optimized image and OS version under Boot disk image in the Marketplace tab.

    The following special OS images with NVIDIA drivers pre-installed are available for VMs with GPUs:

    Intel Broadwell with NVIDIA® Tesla® V100 and Intel Cascade Lake with NVIDIA® Tesla® V100
    Intel Ice Lake with NVIDIA® Tesla® T4
    Intel Ice Lake with T4i
    AMD EPYC™ with NVIDIA® Ampere® A100

    For cluster mode support:

    We recommend using a standard Yandex Cloud image. You can also manually install the drivers on another standard image or create a custom image with pre-installed drivers.

  6. Under Location, select an availability zone for your VM.

  7. Optionally, configure the boot disk under Disks and file storages:

    • Select the disk type.

    • Specify the required disk size.

    • Optionally, to encrypt a boot disk or a secondary disk, under Disks and file storages, click to the right of the disk name and configure encryption parameters for the disk:

      • Select Encrypted disk.
      • In the KMS key field, select the key you want to use to encrypt the disk. To create a new key, click Create.

      To create an encrypted disk, you need the kms.keys.user role or higher.

      Warning

      You can specify encryption settings only when creating a disk. You cannot disable or change disk encryption. You also cannot enable encryption for an existing disk.

      If you deactivate the key used to encrypt a disk, image, or snapshot, access to the data will be suspended until you reactivate the key.

      Alert

      If you destroy the key or its version used to encrypt a disk, image, or snapshot, access to the data will be irrevocably lost. For details, see Destroying key versions.

      If you are creating a VM from an existing boot disk, update the settings of that disk in the Custom tab under Boot disk image at the top of the form.

  8. Optionally, add a secondary disk:

    • Under Disks and file storages, click Add.

    • In the window that opens, select Disk. You can select an existing disk or create a new one, either empty or from a snapshot or image.

      For example, to create a new empty disk:

      • Select Create new disk.
      • In the Contents field, select Empty.
      • Enter a name for the disk.
      • Select the disk type.
      • Specify the required disk and block size.
      • Optionally, enable Additional in the Delete along with the virtual machine field if you need this disk automatically deleted when deleting the VM.
      • Click Add disk.

  9. Optionally, connect a file storage:

    • Under Disks and file storages, click Add.

      • In the window that opens, select File storage and choose the storage you want to connect from the list.

      • Click Add file storage.

  10. Under Computing resources:

    • Go to the GPU tab.

    • Select one of these platforms:

      • Intel Broadwell with NVIDIA® Tesla® V100
      • Intel Cascade Lake with NVIDIA® Tesla® V100
      • AMD EPYC™ with NVIDIA® Ampere® A100
      • Intel Ice Lake with NVIDIA® Tesla® T4
      • Intel Ice Lake with T4i

    • Select one of the available configurations with the required number of GPUs, vCPUs, and amount of RAM.

  11. Under Network settings:

    • In the Subnet field, enter the ID of a subnet in the new VM’s availability zone. Alternatively, select a cloud network from the list.

      • Each network must have at least one subnet. If your network has no subnets, create one by selecting Create subnet.

      • If you do not have a network, click Create network to create one:

        • In the window that opens, specify the network name and select the folder to host the network.
        • Optionally, enable the Create subnets setting to automatically create subnets in all availability zones.
        • Click Create network.

    • In the Public IP address field, select an IP address assignment method:

      • Auto: To assign a random IP address from the Yandex Cloud IP address pool. In this case, you can enable DDoS protection using the option below.
      • List: To select a public IP address from the list of previously reserved static addresses. For more information, see Converting a dynamic public IP address to static.
      • No address: Not to assign a public IP address.

    • Select the relevant security groups. If you leave this field empty, the default security group will be assigned to the VM.

    • Expand Additional and select a method for assigning internal addresses in the Internal IPv4 address field:

      • Auto: To assign a random IP address from the pool of IP addresses available in the selected subnet.
      • Manual: To manually assign a private IP address to the VM.
      • Enable DDoS protection, if required. The option is available if you previously selected the automatic IP assignment method in the public address settings.

    • Optionally, create records for your VM in the DNS zone:

      • Expand DNS settings for internal addresses and click Add record.
      • Specify a zone, FQDN, and TTL for the record. When setting the FQDN, you can enable Detect automatically for the zone.
        You can add multiple records to internal DNS zones. For more information, see Cloud DNS integration with Compute Cloud.
      • To create another record, click Add record.

    If you want to add another network interface to your VM, click Add network interface and repeat the settings from this step for the new interface. You can add up to eight network interfaces to a single VM.

  12. Under Access:

    • Select Access by OS Login to connect and manage access to the new VM using OS Login in Yandex Cloud Organization.

      With OS Login, you can connect to VMs using SSH keys and SSH certificates via a standard SSH client or the Yandex Cloud CLI. OS Login enables rotating the SSH keys used to access VMs, providing the most secure access option.

    • If you prefer not to use OS Login, select SSH key and specify the following VM access data:

      • Under Login, enter a username.

        Alert

        Do not use root or other reserved usernames. To perform operations requiring root privileges, use the sudo command.

      • In the SSH key field, select the SSH key saved in your organization user profile.

        If there are no saved SSH keys in your profile, or you want to add a new key:

        • Click Add key.
        • Enter a name for the SSH key.
        • Upload or paste the contents of the public key file. You need to create a key pair for the SSH connection to a VM yourself.
        • Click Add.

        The SSH key will be added to your organization user profile.

        If users cannot add SSH keys to their profiles in the organization, the added public SSH key will only be saved to the user profile of the VM being created.

    If you want to add multiple users with SSH keys to the VM at the same time, specify these users' data under Metadata. You can also use metadata to install additional software on a VM when creating it.

    In public Linux images provided by Yandex Cloud, the functionality of connecting over SSH using login and password is disabled by default.

  13. Under General information, specify the VM name:

    • It must be from 2 to 63 characters long.
    • It may contain lowercase Latin letters, numbers, and hyphens.
    • It must start with a letter and cannot end with a hyphen.

    Note

    The VM name is used to generate an internal FQDN, which is set only once, when you create the VM. If the internal FQDN is important to you, make sure to choose an appropriate name for your VM.

  14. Under Additional:

  15. Click Create VM.

The VM will appear in the list.

If you do not have the Yandex Cloud CLI yet, install and initialize it.

The folder specified when creating the CLI profile is used by default. To change the default folder, use the yc config set folder-id <folder_ID> command. You can specify a different folder using the --folder-name or --folder-id parameter.

  1. See the description of the CLI command for creating a VM:

    yc compute instance create --help

  2. Prepare a key pair (public and private keys) for SSH access to the VM.

  3. Select a public image.

    To get a list of available images, run the following command:

    yc compute image list --folder-id standard-images

    Result:

    +----------------------+----------------------------------+-----------------------+----------------------+--------+
|          ID          |               NAME               |         FAMILY        |     PRODUCT IDS      | STATUS |
+----------------------+----------------------------------+-----------------------+----------------------+--------+
...
| fdv7ooobjfl3******** | windows-2016-gvlk-gpu-1548913814 | windows-2016-gvlk-gpu | dqnnc72gj2is******** | READY  |
| fdv4f5kv5cvf******** | ubuntu-1604-lts-gpu-1549457823   | ubuntu-1604-lts-gpu   | dqnnb6dc7640******** | READY  |
...
+----------------------+----------------------------------+-----------------------+----------------------+--------+

    The following special OS images with NVIDIA drivers pre-installed are available for VMs with GPUs:

    Intel Broadwell with NVIDIA® Tesla® V100 and Intel Cascade Lake with NVIDIA® Tesla® V100
    Intel Ice Lake with NVIDIA® Tesla® T4
    Intel Ice Lake with T4i
    AMD EPYC™ with NVIDIA® Ampere® A100

    For cluster mode support:

    We recommend using a standard Yandex Cloud image. You can also manually install the drivers on another standard image or create a custom image with pre-installed drivers.

  4. Create a VM in the default folder:

    yc compute instance create \
  --name gpu-instance \
  --zone ru-central1-a \
  --platform=gpu-standard-v3 \
  --cores=8 \
  --memory=96 \
  --gpus=1 \
  --network-interface subnet-name=default-ru-central1-a,nat-ip-version=ipv4 \
  --create-boot-disk image-folder-id=standard-images,image-family=ubuntu-1604-lts-gpu,kms-key-id=<key_ID> \
  --ssh-key ~/.ssh/id_ed25519.pub

    Where:

    • --name: VM name.

      Note

      The VM name is used to generate an internal FQDN, which is set only once, when you create the VM. If the internal FQDN is important to you, make sure to choose an appropriate name for your VM.

    • --zone: Availability zone.

      You can create VMs based on Intel Broadwell with NVIDIA® Tesla® V100, Intel Cascade Lake with NVIDIA® Tesla® V100, and AMD EPYC™ with NVIDIA® Ampere® A100 in the ru-central1-a and ru-central1-b availability zones.

    • --platform: Platform ID:

      • gpu-standard-v1 for Intel Broadwell with NVIDIA® Tesla® V100.
      • gpu-standard-v2 for Intel Cascade Lake with NVIDIA® Tesla® V100.
      • gpu-standard-v3 for AMD EPYC™ with NVIDIA® Ampere® A100.
      • standard-v3-t4 for Intel Ice Lake with NVIDIA® Tesla® T4.
      • standard-v3-t4i for Intel Ice Lake with T4i.

    • --cores: Number of vCPUs.

    • --memory: Amount of RAM.

    • --gpus: Number of GPUs.

    • --preemptible: For a preemptible VM.

    • --network-interface: VM network interface settings:

      • subnet-name: Name of the selected subnet.
      • nat-ip-version=ipv4: Public IP address. To create a VM without a public IP address, omit this parameter.

      If you want to add multiple network interfaces to your VM, specify the --network-interface parameter as many times as you need. You can add up to eight network interfaces to a single VM.

    • --create-boot-disk: OS image.

      • image-family: Image family, e.g., ubuntu-1604-lts-gpu. This option allows you to install the latest version of the OS from the specified family.

      • kms-key-id: ID of the KMS symmetric key to create en encrypted boot disk. This is an optional parameter.

        To create an encrypted disk, you need the kms.keys.user role or higher.

        Warning

        You can specify encryption settings only when creating a disk. You cannot disable or change disk encryption. You also cannot enable encryption for an existing disk.

        If you deactivate the key used to encrypt a disk, image, or snapshot, access to the data will be suspended until you reactivate the key.

        Alert

        If you destroy the key or its version used to encrypt a disk, image, or snapshot, access to the data will be irrevocably lost. For details, see Destroying key versions.

    • --ssh-key: Path to the file with the public SSH key. The VM will automatically create a user named yc-user for this key.

    Get a description of the created VM:

    yc compute instance get --full gpu-instance

    Result:

    name: gpu-instance
zone_id: ru-central1-a
platform_id: gpu-standard-v3
resources:
  memory: "103079215104"
  cores: "8"
  core_fraction: "100"
  gpus: "1"
status: RUNNING
...

If you do not have Terraform yet, install it and configure its Yandex Cloud provider.

  1. In the configuration file, define the parameters of the resources you want to create:

    resource "yandex_compute_disk" "boot-disk" {
  name     = "<disk_name>"
  type     = "<disk_type>"
  zone     = "<availability_zone>"
  size     = "<disk_size>"
  image_id = "<image_ID>"
}

resource "yandex_compute_instance" "vm-1" {
  name                      = "vm-with-gpu"
  allow_stopping_for_update = true
  platform_id               = "standard-v3"
  zone                      = "<availability_zone>"

  resources {
    cores  = <number_of_vCPUs>
    memory = <RAM_in_GB>
    gpus   = <number_of_GPUs>
  }

  boot_disk {
    disk_id = yandex_compute_disk.boot-disk.id
  }

  network_interface {
    subnet_id = "${yandex_vpc_subnet.subnet-1.id}"
    nat       = true
  }

  metadata = {
    ssh-keys = "<username>:<SSH_key_contents>"
  }
}

resource "yandex_vpc_network" "network-1" {
  name = "network1"
}

resource "yandex_vpc_subnet" "subnet-1" {
  name       = "subnet1"
  zone       = "<availability_zone>"
  network_id = "${yandex_vpc_network.network-1.id}"
}

    Where:

    • yandex_compute_disk: Boot disk description:

    • yandex_compute_instance: VM description:

      • name: VM name.

      • allow_stopping_for_update: Permission to stop the VM for updates. Set to true if you plan to change your VM's network settings or computing resources using Terraform. The default value is false.

      • platform_id: Platform ID:

      • zone: Availability zone the VM will reside in.

        You can create VMs based on Intel Broadwell with NVIDIA® Tesla® V100, Intel Cascade Lake with NVIDIA® Tesla® V100, and AMD EPYC™ with NVIDIA® Ampere® A100 in the ru-central1-a and ru-central1-b availability zones.

      • platform_id: Platform ID:

        • gpu-standard-v1 for Intel Broadwell with NVIDIA® Tesla® V100.
        • gpu-standard-v2 for Intel Cascade Lake with NVIDIA® Tesla® V100.
        • gpu-standard-v3 for AMD EPYC™ with NVIDIA® Ampere® A100.
        • standard-v3-t4 for Intel Ice Lake with NVIDIA® Tesla® T4.
        • standard-v3-t4i for Intel Ice Lake with T4i.

      • resources: Number of vCPUs and amount of RAM available to the VM. The values must match the selected platform.

      • boot_disk: Boot disk settings. Specify the disk ID.

      • network_interface: VM network interface settings. Specify the ID of the selected subnet. To automatically assign a public IP address to the VM, set nat = true.

        If you want to add multiple network interfaces to your VM, specify the network_interface section as many times as you need. You can add up to eight network interfaces to a single VM.

      • metadata: In metadata, provide the public key for accessing the VM via SSH. For more information, see VM metadata.

    • yandex_vpc_network: Cloud network description.

    • yandex_vpc_subnet: Description of the subnet to connect your VM to.

    Note

    If you already have suitable resources, such as a cloud network and subnet, you do not need to redefine them. Specify their names and IDs in the appropriate parameters.

    For more information about the resources you can create with Terraform, see the relevant provider documentation.

  2. Create the resources:

    1. In the terminal, change to the folder where you edited the configuration file.

    2. Make sure the configuration file is correct using the command:

      terraform validate

      If the configuration is correct, the following message is returned:

      Success! The configuration is valid.

    3. Run the command:

      terraform plan

      The terminal will display a list of resources with parameters. No changes are made at this step. If the configuration contains errors, Terraform will point them out.

    4. Apply the configuration changes:

      terraform apply

    5. Confirm the changes: type yes in the terminal and press Enter.

    This will create all the resources you need in the specified folder. You can check the new resources and their settings using the management console.

To create a VM, use the create REST API method for the Instance resource or the InstanceService/Create gRPC API call.

When a VM is created, it is assigned an IP address and hostname (FQDN). This data can be used for SSH access.

You can make a public IP address static. For more information, see Making a VM public IP address static.

See also

  • Learn how to update the VM configuration here.
Previous
Creating a VM instance with access to a Yandex Lockbox secret
Next
Making a VM preemptible