Runpod is the managed middle of GPU rental — per-second billing, pods and serverless, Community vs Secure cloud. A 4090 is ~$0.34/hr, an H100 from ~$1.99/hr (24 Sep 2026). Here's what it really costs.
This post contains affiliate links. If you buy through them, Aliteq may earn a commission — at no extra cost to you. Prices verified at publish time.
Share
Runpod is the managed middle of the GPU-rental market: polished, template-driven, billed per second with no egress fees, and offering both classic pods and serverless. It isn't the absolute cheapest — Vast.ai's spot marketplace undercuts it — but as of 24 Sep 2026 an RTX 4090 is about $0.34/hour, an H100 from $1.99/hour, and an A100 80GB $1.19/hour, for a far more predictable experience. Here's what Runpod really costs — the billing model, Secure vs Community cloud, and serverless — and who should skip it. Part of our cloud-GPU pricing pillar.
What Runpod costs right now
Runpod bills per second while a pod runs and charges nothing for data transfer, which keeps the real bill close to the sticker price. It splits into two tiers: Community Cloud runs on vetted third-party hosts and is cheaper, while Secure Cloud runs in RunPod's own data centres on single-tenant hardware at a real premium. An RTX 4090, for example, is about $0.34/hr on Community versus $0.69/hr on Secure. The live figures below are from our tracker, captured 24 Sep 2026 (Vast spot shown for context).
Runpod rate · $/GPU-hour · captured 24 Sep 2026
RTX 4090 · 24GB
Runpod on-demand
$0.34
Vast.ai spot (from)
$0.14
A100 · 80GB
Runpod on-demand
$1.19
Vast.ai spot (from)
$0.47
H100 · 80GB (PCIe)
Runpod on-demand
$1.99
Vast.ai spot (from)
$1.79
H200 · ~141GB
Runpod on-demand
$3.59
Vast.ai spot (from)
$2.63
Runpod on-demand
Vast.ai spot (from)
RTX 4090 · 24GB
$0.34
$0.14
A100 · 80GB
$1.19
$0.47
H100 · 80GB (PCIe)
$1.99
$1.79
H200 · ~141GB
$3.59
$2.63
See the live Runpod rate for your cardReferral link
A pod is a GPU you rent by the hour (on-demand or spot) that stays yours until you stop it — the right choice for development, training, or a steadily-busy service. Serverless flips that: it spins a worker up per request, bills per second, and scales to zero when idle, with fast cold starts (Runpod advertises sub-200 milliseconds) that make it popular for inference APIs. The honest catch is that serverless costs more per active hour than a pod, so it wins only when your GPU would otherwise sit idle — we work through exactly when in our Runpod serverless pricing guide.
Secure vs Community: which to pick
Community Cloud is the value pick — third-party hosts at a lower rate, fine for development and non-sensitive inference. Secure Cloud is the one to choose when compliance or guaranteed uptime matters: it runs on RunPod-operated Tier 3/4 data centres with single-tenant hardware and holds SOC 2 Type II certification (achieved October 2025), which makes it viable for regulated workloads. If your data is sensitive, pay the Secure premium; if it isn't, Community saves you money.
Referral link
Rent a Runpod GPU, per-second billed
A managed, predictable box with templates and no egress fees — RTX 4090 ~$0.34/hr, H100 from ~$1.99/hr (24 Sep 2026). Pick Community to save or Secure for compliance.
Referral link — we may earn a commission at no cost to you. Prices on our compare page are the provider's live figures, cheapest first; this never changes the ranking.
Frequently asked
Is Runpod cheaper than Vast.ai?
Usually not on raw rate — Vast.ai's spot marketplace has the lowest prices (a 4090 from about $0.14/hr versus $0.34/hr on Runpod, 24 Sep 2026). Runpod's value is the managed, predictable experience, per-second billing with no egress fees, and serverless. If the absolute lowest rate is all that matters and you'll vet hosts, Vast wins; for ease and reliability, Runpod is worth the small premium.
What's the difference between Secure Cloud and Community Cloud?
Community Cloud uses vetted third-party hosts and is cheaper; Secure Cloud runs in RunPod's own Tier 3/4 data centres on single-tenant hardware with SOC 2 Type II certification and stronger uptime guarantees. Use Community for development and non-sensitive work to save money, and Secure for regulated or compliance-bound workloads.
Should I use a pod or serverless on Runpod?
Use a pod for steady, predictable load — development, training, or a service that's busy most of the day — because the per-hour pod rate beats serverless once utilisation is high. Use serverless for bursty or low-volume inference, where scale-to-zero means you pay per second instead of for an idle GPU. The break-even is roughly how many hours a day the GPU is genuinely busy.
Is Runpod good for beginners?
Yes — it's one of the easiest GPU clouds to start with, thanks to clear per-second pricing, ready-made templates, and a polished interface. Beginners can launch a pod from a template in minutes without configuring much, then explore serverless once they're comfortable. It's the managed default we recommend to most first-time renters.
Referral link
Managed pods + serverless
Building an app with bursty traffic? Runpod Serverless bills per second and scales to zero, so an idle endpoint costs nothing — check the serverless rate before you deploy.
Runpod is the managed default: not the cheapest rate, but the easiest and most predictable, with serverless as a real edge for app builders. Compare live prices on our cloud-GPU compare page, read the pricing pillar, the Runpod vs Vast vs Lambda comparison, and the Vast.ai review for the cheaper marketplace alternative.
