xpu liveBETA
GPU comparisonGPU providers

GPU CLOUD DIRECTORY

Find your next GPU.

Compare hardware and rental prices.

  • 20 GPUs
  • 5 providers
  • 115 offers
Meet the providers ↗

HARDWARE & PRICING

GPU explorer 20

How prices compare ↗

20 GPUs

Scroll horizontally to compare all specifications and prices ↔

GPU specifications and published rental prices in US dollars per GPU-hour
GPU modelVRAMArchitectureWorkloadsProvidersListed from USD / GPU-hour
NVIDIA A10 — Reference hardware; rental provider cooling configuration may differ.A10NVIDIA · PCIe↗ 24 GBGDDR6 Ampere Inference and graphics Modal $1.1016Serverless · ModalGPU only · CPU/RAM extra
NVIDIA A100 PCIe 80GB — Reference hardware; rental provider cooling configuration may differ.A100 PCIe 80GBNVIDIA · PCIe↗ 80 GBHBM2e Ampere Training and inference Runpod $1.59On-demand · RunpodPublished list price
NVIDIA A100 SXM 80GB — Reference hardware; rental provider cooling configuration may differ.A100 SXM 80GBNVIDIA · SXM4↗ 80 GBHBM2e Ampere Training and fine-tuning CloudRiftRunpodVast.ai $0.8022On-demand · Vast.aiPublished list price
NVIDIA A40 — Reference hardware; rental provider cooling configuration may differ.A40NVIDIA · PCIe↗ 48 GBGDDR6 Ampere Visualization and AI workloads RunpodVast.ai $0.49On-demand · RunpodPublished list price
SXMPhoto pendingB200NVIDIA · SXM↗ 180 GBHBM3e Blackwell Large-model training and inference ModalRunpodVast.ai $6.2496Serverless · ModalGPU only · CPU/RAM extra
NVIDIA H100 NVL — Reference hardware; rental provider cooling configuration may differ.H100 NVLNVIDIA · PCIe / NVL↗ 94 GBHBM3 Hopper Memory-intensive inference RunpodVast.ai $2.5742On-demand · Vast.aiPublished list price
NVIDIA H100 PCIe — Reference hardware; rental provider cooling configuration may differ.H100 PCIeNVIDIA · PCIe↗ 80 GBHBM2e Hopper Training and inference Runpod $2.89On-demand · RunpodPublished list price
NVIDIA H100 SXM — Reference hardware; rental provider cooling configuration may differ.H100 SXMNVIDIA · SXM5↗ 80 GBHBM3 Hopper Multi-GPU training and inference ModalRunpodVast.ai $1.8963On-demand · Vast.aiPublished list price
SXMPhoto pendingH200NVIDIA · SXM↗ 141 GBHBM3e Hopper Large-model inference and HPC CloudRiftModalRunpodVast.ai $3.5556On-demand · Vast.aiPublished list price
NVIDIA L4 — Reference hardware; rental provider cooling configuration may differ.L4NVIDIA · PCIe↗ 24 GBGDDR6 Ada Lovelace Inference and video processing ModalRunpodVast.ai $0.2689On-demand · Vast.aiPublished list price
NVIDIA L40 — Reference hardware; rental provider cooling configuration may differ.L40NVIDIA · PCIe↗ 48 GBGDDR6 Ada Lovelace Rendering and inference Runpod $0.82On-demand · RunpodPublished list price
NVIDIA L40S — Reference hardware; rental provider cooling configuration may differ.L40SNVIDIA · PCIe↗ 48 GBGDDR6 Ada Lovelace Inference, fine-tuning and rendering CloudRiftModalRunpodVast.ai $0.5356On-demand · Vast.aiPublished list price
NVIDIA RTX 3090 — PNY partner card; rental provider chassis and cooler may differ.RTX 3090NVIDIA · PCIe↗ 24 GBGDDR6X Ampere Development and inference RunpodSaladCloudVast.ai $0.09Serverless · SaladCloudGPU only · CPU/RAM extra
NVIDIA RTX 4090 — PNY partner card; rental provider chassis and cooler may differ.RTX 4090NVIDIA · PCIe↗ 24 GBGDDR6X Ada Lovelace Image generation and inference CloudRiftRunpodSaladCloudVast.ai $0.16Serverless · SaladCloudGPU only · CPU/RAM extra
NVIDIA RTX 5090 — PNY partner card; rental provider chassis and cooler may differ.RTX 5090NVIDIA · PCIe↗ 32 GBGDDR7 Blackwell Image generation and inference CloudRiftRunpodSaladCloudVast.ai $0.25Serverless · SaladCloudGPU only · CPU/RAM extra
NVIDIA RTX 6000 Ada — Reference hardware; rental provider cooling configuration may differ.RTX 6000 AdaNVIDIA · PCIe↗ 48 GBGDDR6 Ada Lovelace Professional rendering and inference Runpod $0.84On-demand · RunpodPublished list price
NVIDIA RTX A5000 — Reference hardware; rental provider cooling configuration may differ.RTX A5000NVIDIA · PCIe↗ 24 GBGDDR6 Ampere Development and rendering RunpodVast.ai $0.1889On-demand · Vast.aiPublished list price
NVIDIA RTX A6000 — Reference hardware; rental provider cooling configuration may differ.RTX A6000NVIDIA · PCIe↗ 48 GBGDDR6 Ampere Visualization and training RunpodVast.ai $0.3896On-demand · Vast.aiPublished list price
NVIDIA RTX PRO 6000 Blackwell — Server Edition; rental provider configuration may differ.RTX PRO 6000 BlackwellNVIDIA · Provider variant; confirm edition↗ 96 GBGDDR7 Blackwell Rendering and large-model inference CloudRiftModalRunpod $2.09On-demand · RunpodPublished list price
NVIDIA T4 — Reference hardware; rental provider cooling configuration may differ.T4NVIDIA · PCIe↗ 16 GBGDDR6 Turing Small-model inference and video Modal $0.5904Serverless · ModalGPU only · CPU/RAM extra
How to compare these prices +

“No reservation” includes on-demand and serverless offers. The rental type and provider are shown next to each price. Serverless prices cover GPU time; CPU, RAM and storage may be extra. Reserved prices require a commitment; spot capacity can be interrupted. Prices are per GPU-hour and follow your filters. A multi-GPU minimum is shown when required. Quote-only and unavailable offers are never used as starting prices. Sources and check dates are available on each GPU page.