Skip to main content

​
Name

nebius ai endpoint create

Creates an endpoint.

This command is deterministic and non-interactive. Use it for scripts, config-driven workflows, CI, and agents.

For human first-time setup, use the guided ai create wizard. Static create commands report validation errors with recovery hints instead of switching into prompt mode.

Exposed ports are reachable over managed URLs (no public IP required), surfaced in the endpoint’s status.public_endpoints:

  • an HTTP port (—container-port 8080, or …/http) is served over https://HOST (gRPC works over the same URL);
  • a TCP port (—container-port 6379/tcp) is served over tls://HOST:443 (connect with a direct-TLS client);
  • a UDP port gets no managed URL.

Example (two HTTP ports plus one TLS-TCP port):
cli ai endpoint create —image IMAGE —parent-id PROJECT_ID <br /> —container-port 8080 —container-port 8081/http —container-port 6379/tcp
-> two https:// URLs and one tls://HOST:443 URL in status.public_endpoints.

​
Synopsis

nebius ai endpoint create
  --args
  --async
  --auth
  --container-command
  --container-port
  --disk-size
  --env
  --env-secret
  --image
  --inject-file
  --name
  --parent-id
  --platform
  --preemptible
  --preset
  --public
  --registry-password
  --registry-secret
  --registry-username
  --shm-size
  --ssh-key
  --subnet-id
  --token
  --token-secret
  --volume
  --working-dir

​
Options

--args (string)

Override container arguments.

--async (bool)

Return immediately after starting the create operation.

--auth (string)

Authentication method, default is none.

A value must be one of:

  • none
  • token

--container-command (string)

Override container entrypoint command.

--container-port (string array)

Port to expose as HOST[:CONTAINER][/http|tcp|udp] (default protocol http; repeatable). For endpoints, each HTTP port is reachable over a managed https:// URL (gRPC works over the same URL) and each TCP port over a tls://HOST:443 URL; UDP ports get no managed URL.

--disk-size (string)

Disk size (e.g., 100Gi, 500Gi, 1Ti), default is 250Gi.

--env (string array)

Environment variable in KEY=VALUE format (repeatable).

--env-secret (string array)

Environment variable from MysteryBox in KEY=SECRET_SELECTOR format (repeatable). SECRET_SELECTOR can be a secret name, secret ID, version ID, or SECRET_ID@VERSION_ID.

--image (string)

Container image reference in the registry/path:tag or registry/path@digest format.

--inject-file (string array)

Inject a local config file into the container at LOCAL_PATH:CONTAINER_PATH (repeatable). CONTAINER_PATH must be an absolute path; the file is read-only inside the container and limited to 64 KiB.

--name (string)

Application name.

--parent-id (string)

Parent ID (if not specified, it will be taken from the CLI profile).

--platform (string)

Platform of compute resources (e.g. gpu-h100-sxm, gpu-l40s-d), default is gpu-h100-sxm in eu-north1 and gpu-h200-sxm elsewhere.

--preemptible (bool)

Use a preemptible VM (cheaper but can be stopped by the platform at any time).

--preset (string)

Preset for selected platform (e.g. 1gpu-16vcpu-200gb), default is minimum available preset for the platform.

--public (bool)

Assign a public IP to the workload (default false). For endpoints this is independent of the managed https:// and tls:// URLs, which are provided regardless of this flag.

--registry-password (string)

Container registry password.

--registry-secret (string)

MysteryBox secret selector with REGISTRY_USERNAME and REGISTRY_PASSWORD payload keys. Selector can be a secret name, secret ID, version ID, or SECRET_ID@VERSION_ID.

--registry-username (string)

Container registry username.

--shm-size (string)

Size of /dev/shm (e.g., 64Mi, 128Mi, 1Gi), default - 16Gi for GPU platforms, 0 for CPU.

--ssh-key (string array)

Configure ssh_authorized_keys for creating VM.

--subnet-id (string)

Network subnet ID.

--token (string)

Token used for endpoint authentication (if —auth=token and neither —token nor —token-secret is set, a random token is generated).

--token-secret (string)

MysteryBox secret selector with AUTH_TOKEN payload key. Selector can be a secret name, secret ID, version ID, or SECRET_ID@VERSION_ID.

--volume (string array)

Volume mount in SOURCE:CONTAINER_PATH[:MODE] or s3://BUCKET:/container_path[:MODE[:PROFILE]] format. MODE: rw, ro (default: rw). PROFILE: AWS credentials profile (default: ‘default’). For S3 MysteryBox auth use PROFILE@SECRET_SELECTOR, where SECRET_SELECTOR can be a secret name, secret ID, version ID, or SECRET_ID@VERSION_ID (repeatable).

--working-dir (string)

Working directory (absolute path).

​
Global Options

-h, --help (bool)

Show this message.

-p, --profile (string)

Set a profile for interacting with the cloud.

--format (string)

Output format. Supported values: yaml|json|jsonpath|table|text.

-f, --file (string)

Input file. For ‘update’ commands automatically set —full=true.

-c, --config (string)

Provide path to config file.

--debug (bool)

Enable debug logs.

-I, --impersonate-service-account-id (string)

Impersonate into the service account and use its token for a command.

--color (bool)

Enable colored output.

--no-browser (bool)

Do not open browser automatically on auth.

--insecure (bool)

Disable transport security.

--auth-timeout (duration: 2h30m10s)

Set the timeout for the request including authentication process, default is 15m0s.

--per-retry-timeout (duration: 2h30m10s)

Set the timeout for each retry attempt, default is 20s.

--retries (uint)

Set the number of retry attempts, 1 is disable retries, default is 3.

--timeout (duration: 2h30m10s)

Set the timeout for the main request, default is 1m0s.

--no-check-update (bool)

Suppress check for updates.

--no-progress (bool)

Suppress progress indicators and spinners.

Auto generated on 7-Jul-2026