CH·02CLI reference

gpu instances list

List your GPU instances

Synopsis

Lists GPU instances visible to the authenticated org. By default only non-terminated instances are shown — the live fleet, not the full provision history. Use --status terminated for the history, --status to filter by a specific state, or --all for everything. Output is paginated up to --limit rows (default 50, max 200). On TTY the result is a tabwriter-aligned table; off-TTY it defaults to JSON.

gpu instances list [flags]

Examples

# Table view (default on TTY)
gpu instances list

# JSON for jq pipelines
gpu instances list --output json | jq '.[] | {id, status}'

# Limit results
gpu instances list --limit 10

# Only running instances
gpu instances list --status running

# Full history including terminated instances
gpu instances list --all

An AUTO-TERM column appears when at least one listed instance was launched with --auto-terminate-hours, showing when the platform will terminate it (instances with no limit show -). The column is absent when nothing in the listing has a limit, so the usual fleet view is unchanged. Use --output json for a stable shape (auto_terminate_hours, auto_terminate_at).

The STATUS column can also show unreachable: a running instance whose SSH tunnel GPU.ai has confirmed dead (several consecutive failed probes). It is a display state on top of running — filter with --status running — and the confirmed-dead window is not billed (credited on recovery, or billing closes back to the outage start if the instance is eventually failed). gpu instances get shows the outage start as Unreachable since:.

Options

      --all             Include terminated instances (shorthand for --status=all)
  -h, --help            help for list
      --limit int       Max instances to fetch (1-200) (default 50)
      --status string   Filter by status: allocating | starting | running | stopping | stopped | terminated | error | all (default: all non-terminated; unreachable instances list under running)

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)

Synopsis

Examples

Options

Options inherited from parent commands

SEE ALSO