GPU-Native Cloud Spot Available

RunPod Pricing

Visit Website Official Pricing
Available GPUs
50
Instance Types
51
H100 Price
$2.89/hr
A100 Price
$1.39/hr
Sponsored

Start renting GPUs on RunPod

Estimate cost first
RunPod GPU rental sponsored offer RunPod Launch pods, serverless GPUs, and templates for AI workloads. Start on RunPod

Price Highlights

Cheapest GPU
RTX 3070
$0.130/hr
H100 (per GPU)
$2.89/hr
A100 (per GPU)
$1.39/hr
Best Spot Discount
49% off
V100 SXM2 32GB

About RunPod

Key Features
  • Serverless GPUs
  • Pay-per-second
  • Community cloud
  • Templates
Best For
  • Inference APIs
  • Short-term workloads
  • Cost-conscious users
Considerations
  • Variable availability
  • Community instances less reliable

RunPod Pricing and H100 Cost Planning

RunPod pricing searches are often not just generic GPU pricing. Users compare RunPod H100 rates, Community Cloud versus Secure Cloud, monthly A100 cost, spend limits, small-VRAM pod cost, and whether serverless or pods are the cheaper billing model for a specific workload.

RunPod H100 pricing by cloud type

When checking an H100 row, record whether the capacity is Community Cloud or Secure Cloud, the GPU count, region, and billing unit. Community Cloud can be cheaper for experiments, while Secure Cloud is usually easier to justify for production inference and private data.

RunPod A100 monthly cost

A monthly estimate should multiply the active GPU rate by expected runtime, then add persistent volume, network transfer, stopped-pod storage, and idle time. Do not compare only the hourly GPU headline.

Spend limit and short tests

For first-time tests, set a budget guardrail, start with the smallest VRAM tier that fits, and move up only when memory, latency, or throughput actually becomes the bottleneck.

RunPod serverless vs pods cost example

A pod is usually easier to budget for a long training run because active time is continuous. Serverless can be cheaper for intermittent inference, but cold starts, worker idle time, request concurrency, and storage can change the result. Model one real day of traffic before extrapolating to a month.

RunPod H100 price and throughput check

A lower H100 GPU-hour is not automatically a lower project cost. Compare startup time, sustained throughput, storage, retries, and whether the selected 80GB configuration can hold the model and batch size without repeated out-of-memory failures.

Run these checks before committing to a long RunPod GPU rental.
Cost check Why it matters
Check 1 For short experiments, compare a pod against serverless by expected cold-start tolerance, runtime length, and whether the job needs persistent storage.
Check 2 For a month of A100 work, model active hours, stopped storage, dataset staging, and failed-run retries before comparing RunPod against Lambda, Vast.ai, or hyperscalers.
Check 3 For small 8GB VRAM jobs, verify model memory use first. A cheaper small pod is not a saving if it forces repeated out-of-memory failures or lower-quality settings.
Check 4 For RunPod H100 pricing, capture the exact 80GB row, cloud type, region, GPU count, and availability before comparing it with another provider.
Related: GPU rental cost calculator A100 monthly cost calculator Serverless GPU pricing calculator H100 GPU pricing comparison H100 NVL pricing Vast.ai marketplace comparison All cloud GPU prices

H100 Price Comparison

Compare RunPod's H100 pricing with other gpu-cloud providers.

Provider H100 Price (per GPU) vs RunPod
Fluidstack $2.10/hr -27%
Genesis Cloud $2.19/hr -24%
Paperspace $2.24/hr -22%
Datacrunch (Verda) $2.29/hr -21%
RunPod Current $2.89/hr -
Jarvis Labs $2.99/hr +3%
Lambda Labs $3.29/hr +14%
CoreWeave $4.25/hr +47%

All GPU Pricing

Data Center GPUs 10 instances
GPU VRAM Instance GPUs On-Demand Per GPU Spot Availability
B200 192GB NVIDIA B200 1x $6.79/hr $6.79/hr $6.79/hr -
H200 141GB NVIDIA H200 1x $4.59/hr $4.59/hr $4.59/hr -
H100 80GB NVIDIA H100 80GB HBM3 1x $3.29/hr $3.29/hr $3.29/hr -
H100 PCIe 80GB NVIDIA H100 PCIe 1x $2.89/hr $2.89/hr $2.89/hr -
MI300X 192GB AMD Instinct MI300X OAM 1x $2.39/hr $2.39/hr $2.39/hr -
H100 NVL 93GB NVIDIA H100 NVL 1x $3.19/hr $3.19/hr $3.19/hr -
A100 SXM 80GB NVIDIA A100-SXM4-80GB 1x $1.59/hr $1.59/hr $1.59/hr -
L40S 48GB NVIDIA L40S 1x $0.990/hr $0.990/hr $0.990/hr -
L4 22GB NVIDIA L4 1x $0.490/hr $0.490/hr $0.490/hr -
A40 48GB NVIDIA A40 1x $0.440/hr $0.440/hr Cheapest $0.440/hr -
Consumer GPUs 8 instances
GPU VRAM Instance GPUs On-Demand Per GPU Spot Availability
RTX 4090 24GB NVIDIA GeForce RTX 4090 1x $0.740/hr $0.740/hr $0.740/hr -
RTX 4080 15GB NVIDIA GeForce RTX 4080 1x $0.500/hr $0.500/hr $0.500/hr -
RTX 4080 15GB NVIDIA GeForce RTX 4080 SUPER 1x $0.500/hr $0.500/hr $0.500/hr -
RTX 3090 24GB NVIDIA GeForce RTX 3090 1x $0.500/hr $0.500/hr $0.500/hr -
RTX 3080 10GB NVIDIA GeForce RTX 3080 1x $0.170/hr $0.170/hr Cheapest $0.170/hr -
RTX 3080 Ti 12GB NVIDIA GeForce RTX 3080 Ti 1x $0.180/hr $0.180/hr $0.180/hr -
RTX 3090 Ti 24GB NVIDIA GeForce RTX 3090 Ti 1x $0.460/hr $0.460/hr $0.460/hr -
RTX 4070 Ti 12GB NVIDIA GeForce RTX 4070 Ti 1x $0.500/hr $0.500/hr $0.500/hr -
Workstation GPUs 3 instances
GPU VRAM Instance GPUs On-Demand Per GPU Spot Availability
RTX 6000 Ada 48GB NVIDIA RTX 6000 Ada Generation 1x $0.840/hr $0.840/hr $0.840/hr -
RTX A6000 48GB NVIDIA RTX A6000 1x $0.530/hr $0.530/hr $0.530/hr -
RTX A4000 16GB NVIDIA RTX A4000 1x $0.250/hr $0.250/hr Cheapest $0.250/hr -
Other GPUs 30 instances
GPU VRAM Instance GPUs On-Demand Per GPU Spot Availability
A100 80GB 80GB NVIDIA A100 80GB PCIe 1x $1.39/hr $1.39/hr $1.39/hr -
A30 24GB NVIDIA A30 1x $0.410/hr $0.410/hr $0.210/hr -
RTX 3070 8GB NVIDIA GeForce RTX 3070 1x $0.130/hr $0.130/hr Cheapest $0.130/hr -
RTX 5080 16GB NVIDIA GeForce RTX 5080 1x $0.590/hr $0.590/hr $0.590/hr -
RTX 5090 32GB NVIDIA GeForce RTX 5090 1x $0.990/hr $0.990/hr $0.990/hr -
NVIDIA H200 NVL 143GB NVIDIA H200 NVL 1x $3.39/hr $3.39/hr - -
L40 48GB NVIDIA L40 1x $0.820/hr $0.820/hr $0.820/hr -
RTX 2000 Ada 16GB NVIDIA RTX 2000 Ada Generation 1x $0.240/hr $0.240/hr $0.240/hr -
RTX 4000 Ada 20GB NVIDIA RTX 4000 Ada Generation 1x $0.280/hr $0.280/hr $0.280/hr -
RTX 4000 Ada SFF 20GB NVIDIA RTX 4000 SFF Ada Generation 1x $0.440/hr $0.440/hr $0.440/hr -
RTX 5000 Ada 32GB NVIDIA RTX 5000 Ada Generation 1x $0.830/hr $0.830/hr $0.830/hr -
RTX A2000 6GB NVIDIA RTX A2000 1x $0.500/hr $0.500/hr $0.500/hr -
RTX A4500 20GB NVIDIA RTX A4500 1x $0.250/hr $0.250/hr $0.250/hr -
RTX A5000 24GB NVIDIA RTX A5000 1x $0.270/hr $0.270/hr $0.270/hr -
RTX PRO 6000 MaxQ 96GB NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 1x $0.500/hr $0.500/hr $0.500/hr -
RTX PRO 6000 96GB NVIDIA RTX PRO 6000 Blackwell Server Edition 1x $2.09/hr $2.09/hr $2.09/hr -
RTX PRO 6000 WK 96GB NVIDIA RTX PRO 6000 Blackwell Workstation Edition 1x $1.89/hr $1.89/hr $1.89/hr -
V100 FHHL 16GB Tesla V100-FHHL-16GB 1x $0.190/hr $0.190/hr $0.100/hr -
Tesla V100 16GB Tesla V100-PCIE-16GB 1x $0.190/hr $0.190/hr $0.190/hr -
V100 SXM2 16GB Tesla V100-SXM2-16GB 1x $0.230/hr $0.230/hr $0.230/hr -
V100 SXM2 32GB 32GB Tesla V100-SXM2-32GB 1x $0.490/hr $0.490/hr $0.250/hr -
A100 SXM 40GB 40GB NVIDIA A100-SXM4-40GB 1x $1.00/hr $1.00/hr $1.00/hr -
B300 288GB NVIDIA B300 SXM6 AC 1x $7.39/hr $7.39/hr $7.39/hr -
H200 NVL 143GB NVIDIA H200 NVL 1x $3.79/hr $3.79/hr $3.79/hr -
RTX PRO 4000 24GB NVIDIA RTX PRO 4000 Blackwell 1x $0.570/hr $0.570/hr $0.570/hr -
RTX PRO 4500 32GB NVIDIA RTX PRO 4500 Blackwell 1x $0.720/hr $0.720/hr $0.720/hr -
NVIDIA RTX PRO 4500 Blackwell Server Edition 32GB NVIDIA RTX PRO 4500 Blackwell Server Edition 1x $0.500/hr $0.500/hr $0.500/hr -
RTX PRO 5000 48GB NVIDIA RTX PRO 5000 Blackwell 1x $0.960/hr $0.960/hr $0.960/hr -
PRO 6000 MIG 24GB 24GB NVIDIA RTX PRO 6000 Blackwell Server Edition MIG 1g.24gb 1x $0.490/hr $0.490/hr $0.490/hr -
PRO 6000 MIG 48GB 48GB NVIDIA RTX PRO 6000 Blackwell Server Edition MIG 2g.48gb 1x $1.00/hr $1.00/hr $1.00/hr -

Similar Providers

Other gpu-cloud providers you might consider.

Genesis Cloud
0 GPUs
CoreWeave
12 GPUs
from $0.390/hr
Datacrunch (Verda)
9 GPUs
from $0.140/hr
Fluidstack
5 GPUs
from $1.30/hr
Genesis Cloud
4 GPUs
from $0.250/hr
Jarvis Labs
6 GPUs
from $0.490/hr
View All Providers →

Frequently Asked Questions about RunPod

RunPod H100 cost depends on the exact 80GB row, Community or Secure Cloud, GPU count, region, availability, and billing state. Use the live table on this page for the current reference, then verify the selected pod before starting a long job.

Compare the per-GPU rate, cloud type, startup time, sustained throughput, storage, network transfer, and expected interruption or host-quality risk. A cheaper headline rate may not be cheaper when the job needs more retries or a larger host.

Multiply the selected A100 hourly rate by expected active hours, then add persistent storage, network transfer, idle time, and failed-run retries. A 24/7 workload should be compared separately from a few hours of daily development.

Community Cloud is marketplace-style capacity that can be cheaper but varies by host. Secure Cloud is better for production-oriented workloads that need more predictable infrastructure and data handling.

RunPod users should configure budget controls and monitor active pods, volumes, and serverless usage. This page cannot enforce billing limits, so always confirm current account controls inside RunPod before running long jobs.

It can be enough for small notebooks, lightweight inference, and simple demos. Larger LLMs, image generation at higher settings, or batch jobs may need 16GB, 24GB, 40GB, or 80GB GPUs.

Do not assume a stopped pod has zero cost. Persistent volumes and retained storage may continue to generate charges, so include storage duration and cleanup timing in any monthly estimate and verify current terms on RunPod.

RunPod charges $2.89/hr per H100 GPU for on-demand instances. Prices may vary based on instance type, region, and commitment length. Check their official pricing page for the most current rates.

RunPod is well-suited for AI/ML training with competitive pricing and good GPU availability. They specialize in GPU compute and often have better availability for high-demand GPUs like H100 and A100.

Check RunPod's pricing page for spot or preemptible instance options. These can offer 50-90% savings but come with the risk of interruption.

RunPod offers a good balance of price and reliability compared to hyperscalers. They typically have better GPU availability and lower prices than AWS/Azure/GCP, while being more reliable than marketplace providers.

Most cloud GPU providers including RunPod accept credit cards for pay-as-you-go usage. Enterprise customers may have options for invoicing and purchase orders. Check their website for specific payment methods and billing options.