RunPod pricing explained: what RunPod charges in 2026

18 October 2026

This post contains affiliate links. If you sign up via our links we may earn a small commission at no extra cost to you. We run ComputeClarity.

This page explains what RunPod charges for GPUs. Every figure comes from RunPod's own pages, read on 18 October 2026, and RunPod's pricing page was showing "Updated October 8, 2026" at the time. RunPod changes its prices, so treat this as a dated snapshot and confirm on the pricing page before you commit. We run ComputeClarity and earn a commission if you sign up to RunPod through our link.

What RunPod lists for Pods

Pods are RunPod's rented GPU machines. The pricing page lists one price per GPU per hour. It does not say whether each price belongs to Secure Cloud or Community Cloud, so check which tier you are deploying to when you start a Pod. These were the listed rates on 18 October 2026:

NVIDIA B300, 288 GB: $8.99 per hour.

NVIDIA B200, 180 GB: $7.99 per hour.

NVIDIA H200, 141 GB: $5.29 per hour.

NVIDIA H100 SXM, 80 GB: $3.99 per hour.

NVIDIA H100 NVL, 94 GB: $3.19 per hour.

NVIDIA H100 PCIe, 80 GB: $2.89 per hour.

NVIDIA A100 SXM, 80 GB: $1.79 per hour.

NVIDIA A100 PCIe, 80 GB: $1.79 per hour.

NVIDIA L40S, 48 GB: $1.09 per hour.

NVIDIA RTX 4090, 24 GB: $0.89 per hour.

NVIDIA RTX 3090, 24 GB: $0.50 per hour.

The page also lists the RTX Pro 6000, RTX 5090, L40, L4, A40, RTX A6000, RTX A5000 and other cards. Compare the memory as well as the name. The H100 appears three times, as SXM, NVL and PCIe, and the prices differ by $1.10 per hour between the cheapest and the dearest.

Working out a job cost

The cost of a job is the rate, times the number of GPUs, times the hours the Pod is running. As an illustration using the H100 SXM rate above, one GPU for 40 hours is 40 x $3.99, which is $159.60. Eight H100 PCIe GPUs for 10 hours is 8 x 10 x $2.89, which is $231.20. Add storage and any time the Pod sits idle. You can compare stored rates across providers in the ComputeClarity calculator.

Serverless

RunPod lists Serverless rates separately, per hour. On 18 October 2026 the page showed 80 GB H100 at $4.79, A100 at $2.72, 24 GB RTX 4090 at $1.10 and the 16 GB group at $0.58. For the same GPU class the Serverless hourly rate is higher than the Pod rate, so compare it with how often your workload is actually busy. The page did not show per second Serverless prices, so check how idle workers are billed before you rely on it for an always on API.

Clusters and reserved capacity

The page lists Clusters at $6.35 per hour for H200 SXM and $2.15 per hour for A100 SXM, scaling up to 64 GPUs. L40S, H100 SXM and B200 clusters are listed as contact sales. Reserved Clusters for 1, 3, 6 and 12 months or longer are also contact sales, with no figures shown.

Storage

These storage rates were listed on 18 October 2026:

Container disk: $0.10 per GB per month. RunPod's docs say it is temporary and erased when the Pod stops.

Volume disk: $0.10 per GB per month while running and $0.20 per GB per month when idle.

Network volume: $0.07 per GB per month under 1 TB and $0.05 over 1 TB. High performance network storage is $0.14 per GB per month.

As an illustration, a 500 GB volume disk costs 500 x $0.10, which is $50 a month while the Pod runs and 500 x $0.20, which is $100 a month when stopped. The same 500 GB on a standard network volume is 500 x $0.07, which is $35 a month. Storage charges carry on for stopped Pods, so delete what you no longer need.

How it bills

RunPod's docs say Pods are billed by the second for compute and storage, with no fees for data ingress or egress. Network volumes are billed hourly. The same docs say a deployment needs at least one hour of credits for the chosen configuration.

Savings plans

RunPod's docs describe savings plans for 3 or 6 months, paid upfront, covering GPU compute only. They apply to your next Pod of the same GPU type, and they are non-refundable with fixed expiry dates. Only buy one if you are confident of steady use.

What happens when credit runs out

Per RunPod's docs, Pods stop automatically when your balance reaches $0. A Pod with a network volume is stopped and its data is kept. A Pod without one is terminated, and the docs say its data cannot be recovered. Turn on auto pay or low balance notifications, and keep important data on a network volume or outside RunPod. The docs also state a default spend limit of $80 per hour across all resources, which you can ask support to raise.

What to check before you start

Match the GPU and its memory to your workload. Check which cloud tier you are deploying to. Work out storage as well as the hourly rate. Run a short test job first, and keep a second provider as a fallback if capacity is tight.

To see how RunPod compares with a marketplace, read Vast.ai pricing explained and RunPod vs Vast.ai. For other options, read RunPod alternatives and GPU cloud providers compared.