CH·02CLI reference

gpu dedicated-ip

Show the dedicated public IP add-on: prices, ports, and availability

Show the dedicated public IP add-on: prices, ports, and availability

Synopsis

Shows the dedicated public IPv4 add-on for instances. The output covers the hourly fee, the transfer price and cap, which inbound TCP ports can be forwarded, how many addresses your organization holds against its limit, and whether one is free right now.

A dedicated IP is a public address owned by GPU.ai (never the supplier's). Inbound TCP on the chosen ports is forwarded, raw and without TLS termination, to the same port on the instance's loopback, so your own certificates (for example from Let's Encrypt) work end to end. Your service must listen on that port inside the instance. Port 22 is never forwarded, and SSH stays on the instance's normal SSH command. Outbound traffic is unchanged.

Attach one at launch with gpu instances create --dedicated-ip (ports 80 and 443) or --dedicated-ip-ports 80,443,8000. It cannot be added later, and it is released automatically when the instance terminates or fails.

The add-on is off by default. On a deployment that does not offer it this command fails with dedicated IPs are not available on this deployment.

gpu dedicated-ip [flags]

Output

Hourly fee         $0.01/hr while attached
Transfer           $0.12/GB, inbound and outbound
Transfer cap       500 GB per address, then forwarding is suspended
Default ports      80, 443
Allowed ports      80,443
Max ports          8
Your organization  0 of 2 in use
Available now      3

A dedicated IP is for serving your own workload (web apps, APIs, model endpoints). ...
  • Hourly fee is charged from the moment the instance becomes running until the address is released.
  • Transfer counts bytes in both directions. When an address reaches the transfer cap, forwarding is suspended so a traffic spike cannot run up an unbounded bill, and gpu instances get shows suspended: transfer_cap.
  • Your organization is how many addresses you hold now against your limit. A launch beyond the limit fails with dedicated_ip_limit.
  • Available now is how many addresses are free. When it is 0 a launch fails with dedicated_ip_unavailable.
Both charges are billed on top of the instance's hourly price, and the rates are locked when the instance launches. Use --output json for the raw GET /v1/dedicated-ip response. See the API guide for DNS, TLS, release, and the acceptable use policy.

Examples

# Prices, ports, and how many you can still attach
gpu dedicated-ip

# The same as JSON
gpu dedicated-ip --output json

# Launch with one
gpu instances create -t h100_sxm --dedicated-ip

Options

  -h, --help   help for dedicated-ip

Options inherited from parent commands

      --api-base string   API base URL (override with GPUAI_API_BASE env) (default "https://api.gpu.ai/v1")
      --debug             Enable debug logging to stderr
  -o, --output string     Output format: table|json (default table on TTY, json otherwise)

SEE ALSO

← The gpu CLI