RunPod Auto-Pay Balance-Triggered Top-Up | Generated by AI

Home 2026.06

Question: How does RunPod’s auto-pay (automatic billing) work, based on the settings shown?

Answer:

It’s a balance-triggered top-up. You prepay RunPod in credits (not postpaid invoicing), and auto-pay keeps that prepaid balance from hitting zero while your pods/endpoints are burning it down.

The two fields map directly to a threshold + reload pair:

The rate limit matters: billed max once per hour. So even if you’re draining credits fast (e.g., a multi-GPU pod at several $/hr), auto-pay won’t fire repeatedly within the same hour — it tops up $25, waits up to an hour, and only re-triggers if you’re still below $10 at the next eligible window.

Concrete walkthrough with your numbers:

balance = $12
running pod burns $4/hr

t=0h   balance $12  → above $10, no charge
t=0.5h balance $10  → at threshold
t=0.6h balance $9.20 → below $10 → CHARGE $25 → balance $34.20
t=1.6h balance $30.20 (1hr later, still draining) → above $10, no charge
...continues until balance dips below $10 again, then next charge eligible

The edge case to watch: if your burn rate exceeds $25/hr, auto-pay can’t keep up. A single MI300X-class or 8×GPU pod can burn faster than $25/hr, and since reloads are capped at once/hour, your balance can still hit $0 mid-hour and your pod gets stopped/terminated before the next reload window. For high-burn workloads, either raise the “add” amount well above your hourly burn, or keep a manual buffer.

Practical config for your usage (AMD Dev Cloud ~$2/hr equivalent, occasional bigger pods):

One more thing: auto-pay is about prepaid credit, so there’s no surprise end-of-month bill — your exposure per hour is bounded by the reload amount. That’s a nice property if you want a hard-ish ceiling on accidental spend (forgot to stop a pod) — though a runaway pod will still drain and re-trigger hour after hour, so set spend limits / pod time limits too if you want a true cap.


Back Donate