Pod statements

A statement lists the charges for one pod. It stays available after you remove the pod.

Where to find it

All pods in a period

Open Billing and scroll to Pods in this period. Expand a pod to see its charges.

One pod

In Your Pods, open the pod's menu and click Billing statement.

Removed pods

In Your Pods, click Statements for removed pods.

Pick a period

Show any start and end date

1. Open Billing.
2. Click the date field next to the period arrows.
3. Pick the start day, then the end day.

The chart and Pods in this period switch to that range. The arrows still step one month back or forward.

What a statement shows

• Total: what your balance was charged for the pod.
• Time billed: the total time charged.
• Price: the hourly price. Two prices mean it changed during the rental.
• Rented / Removed: start and end of the rental, and why it ended.
• Ledger rows: one or more rows per UTC day, oldest first.

Pods are charged every 5 minutes.

For agents and automation: API

Both routes are under https://lium.io/api and accept an X-API-Key header:

• GET /billing/statement?start_day=YYYY-MM-DD&end_day=YYYY-MM-DD: every pod in the period.
• GET /pods/{id}/statement: one pod, before or after removal.

Run record

A run record is one page about a pod you can keep to repeat or cite the run: what it ran on, when, and what it cost. Like the statement, it stays available after the pod is removed.

Field	Meaning
pod_id, pod_name	The pod's ID and name.
removed	true once the pod is removed.
executor_id	The node the pod ran on last.
gpu_name, gpu_count	GPU model and count.
image	The image reference the container was started from. It ends in @sha256:… when the template pins a digest; a tag alone can point at a different image later, so pin a digest in the template when the exact image matters.
launch_snapshot.template_id	The template row the container was started from: usually the pod's own copy of the template you picked, removed with the pod.
launch_snapshot.source_template_id	The template you picked, when template_id is the pod's copy of it. Empty when the pod ran on the template itself (a one-time template) and for a pod whose container came up before this was recorded.
launch_snapshot.image	The image reference recorded when the container came up; image above is this value when the snapshot exists.
launch_snapshot.gpu_driver, launch_snapshot.cuda_driver	NVIDIA driver build and the highest CUDA version it supports, as the node reported them when your container came up.
launch_snapshot.gpu_interconnect	nvlink when every GPU pair on the node is joined by NVLink, pcie when at least one pair is not. Empty on a single-GPU node.
launch_snapshot.recorded_at	When the container came up and these values were recorded.
created_at, removed_at, close_reason	Rental start and end, and why it ended.
total, billed_seconds, price_segments	What you were charged, the same figures as the statement.

If the pod was rebooted, moved to another node, or switched to another template, the record describes the last container that came up. Pods whose container came up before run records existed have no launch_snapshot; image then comes from the rental.

curl https://lium.io/api/pods/<pod_id>/record -H "X-API-Key: $LIUM_API_KEY"

The route answers 404 under the same rule as GET /pods/{id}/statement. GET /version lists "run_record" under features on a backend that serves this route, so a client can tell a backend without it from a pod that is not yours.