RackRateCompare prices ↓
Home/Providers
Cloud provider directory

Compare where you rent GPUs.

Provider type shapes pricing, supply, support, and operational control. Compare the sample catalogue first, then verify the exact region, machine, contract, and capacity behind any published rate.

01Marketplace

Distributed host inventory with flexible pricing and configuration variance.

02GPU cloud

Specialist infrastructure built around accelerator workloads and developer access.

03Hyperscaler

Broad cloud services, formal procurement, and many regional instance families.

Provider scorecards

Compare the operating model, not just the rate.

The cards use the launch sample feed. A provider with fewer rows may have a broader live catalogue than this preview shows, so treat counts as indexed coverage rather than a completeness claim.

MarketplaceUnited States

Vast.ai

A marketplace where independent hosts list GPU capacity across many regions and hardware configurations.

Billing
Per second
Sample coverage
3 offers · 3 regions
GPUs indexed
RTX 4090, A100 80GB, H100
From $0.32/hrReview Vast.ai →
GPU cloudUnited Kingdom

Hyperstack

A GPU-focused cloud with on-demand and reserved capacity across several datacenter regions.

Billing
Per minute
Sample coverage
4 offers · 1 region
GPUs indexed
L40S, A100 80GB, H100, H200
From $0.69/hrReview Hyperstack →
GPU cloudUnited States

Runpod

A developer-oriented GPU platform spanning Pods, serverless inference, and community capacity.

Billing
Per second
Sample coverage
6 offers · 1 region
GPUs indexed
RTX 4090, L40S, A100 80GB, H100, H200, B200
From $0.59/hrReview Runpod →
GPU cloudUnited States

Lambda

A specialist AI cloud offering single-node and multi-node GPU infrastructure.

Billing
Per hour
Sample coverage
1 offers · 1 region
GPUs indexed
H100
From $4.29/hrReview Lambda →
GPU cloudNetherlands

Nebius

An AI infrastructure platform providing GPU virtual machines and managed services.

Billing
Per second
Sample coverage
2 offers · 1 region
GPUs indexed
H200, B200
From $3.85/hrReview Nebius →
HyperscalerUnited States

AWS

A global hyperscale cloud with GPU instances, capacity blocks, and spot purchasing options.

Billing
Per second
Sample coverage
1 offers · 1 region
GPUs indexed
H100
From $3.78/hrReview AWS →
Before you rent GPUs

Six questions for every provider shortlist.

  1. 01

    Is capacity actually bookable?

    A published price does not guarantee stock for your GPU count, region, and start date.

  2. 02

    What is the complete node?

    Confirm CPU, system RAM, local storage, network bandwidth, and accelerator interconnect.

  3. 03

    How is usage metered?

    Per-second billing reduces idle cost, while minimum terms can change the effective hourly price.

  4. 04

    What does data movement cost?

    Estimate ingress, internet egress, cross-zone traffic, and checkpoint transfer separately.

  5. 05

    What happens when a node fails?

    Ask about replacement capacity, credits, persistent storage, and support response.

  6. 06

    Can procurement verify the terms?

    Match the public listing to a quote that names the GPU, quantity, location, term, and exclusions.

Ready to compare offers?

Put every provider on the same unit.

Open the price comparison →