gpu●aiDocs
CH·02CLI reference

gpu billing balance

Show your credit balance, burn rate and runway

Show your credit balance, burn rate and runway

Synopsis

Answers the question you ask before starting work: how much credit is left, and how long does it last at what I am already running?

gpu billing balance reports three figures:

  • Balance — remaining credit on the organization that owns your API key.
  • Burn rate — the committed hourly price of every instance you have live.
It counts an instance from the moment you launch it, so instances still creating, provisioning or booting are included, not only ones that reached running. That is deliberate: it is the same figure the burn cap judges a launch by, and for a runway estimate it is safer to over-count burn than to report a runway you do not have.
  • Runway — balance ÷ burn rate, shown at a resolution that matches its size
(minutes, hours, or days).

There is no runway figure in three cases, and the command says which one:

LineMeans
— (nothing running)No burn to divide by; the balance funds unbounded idle time.
none — out of credit while burningInstances are running and the balance is at or below zero. Act now.
unknown — burn rate unavailableThe burn rate could not be read. Not the same as idle — assume you are burning.

If auto-pay is on, the table says what it will do, because a balance that refills itself has a different meaning from one that does not.

gpu billing balance [flags]

Examples

# How long can I keep running?
gpu billing balance

# Balance in cents, for a script
gpu billing balance --output json | jq .balance_cents

# Refuse to launch when runway is under two hours — and refuse just as firmly
# when the burn rate is unknown, since unknown is not idle.
gpu billing balance --output json | jq -e '
  .burn_cents_per_hour != null
  and (.burn_cents_per_hour == 0 or .balance_cents / .burn_cents_per_hour >= 2)
' >/dev/null || { echo "top up first (or burn rate unavailable)"; exit 1; }

Output

Balance     $42.50
Burn rate   $5.00/hr
Runway      ~8.5h at current burn
Auto-pay    off

JSON mode returns the API response unchanged — balance_cents, balance_dollars, burn_cents_per_hour, and the auto_pay_* fields. The runway is a table-mode convenience and is deliberately not a JSON field, so a script does its own arithmetic against figures the API actually returned.

burn_cents_per_hour is null when the burn rate could not be determined. A script must not treat that as 0 — the difference between "nothing running" and "unknown" is the difference between unbounded runway and none.

You wantCommand
How much credit is leftgpu billing balance
Add creditgpu billing deposit --crypto
What the org has spent, against a capgpu spend-limit
Past consumption, broken downgpu usage

Options

  -h, --help   help for balance

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

  • gpu billing - Check your balance, add funds, and review deposits
  • gpu billing deposit - Add funds to your account balance with a stablecoin deposit
  • gpu spend-limit - View or set the org monthly spending limit and daily spend cap

← The gpu CLI