Best value
cheapest H100
TensorDock
KVM-based marketplace for GPU virtual machines with customizable configurations.
$2.25
Independent GPU cloud price index
Stop overpaying for
GPU cloud compute.
Compare verified hourly pricing, VRAM specs, and instance availability across top GPU cloud providers in one clear index.
Explore
Top picks for H100 right now
Top pick
$2.89/hr
RunPod
Community + secure cloud with broad GPU range and template catalog.
Read review →
Top pick
$3.29/hr
Lambda
Managed GPU cloud and reserved clusters for training.
Read review →In-depth provider reviews
View all reviews →
Top Overall
RunPod
Community and Secure Cloud instances with serverless endpoint options.
Cheapest
Vast.ai
P2P GPU marketplace with lowest cost per hour for budget-conscious workloads.
Enterprise
Lambda Cloud
Developer-first cloud offering high-speed 8x H100 and H200 SXM clusters.
H100 on-demand, at a glance
| Provider | On-demand $/hr | Spot $/hr | Availability |
|---|---|---|---|
| TensorDock Cheapest | $2.25 | n/a | High |
| RunPod | $2.89 | $1.99 | High |
| Lambda | $3.29 | n/a | Medium |
| Lambda | $3.99 | n/a | Medium |
| CoreWeave | $6.16 | $2.46 | High |
Latest articles
DeepSeek V4.1 Flash GPU Requirements: VRAM & Cluster Sizing
Calculate VRAM, memory bandwidth, and GPU cloud cluster sizing for DeepSeek V4.1 Flash (552B MoE) across NVIDIA H200, B200, and AMD MI325X servers.
GPU Cloud vs On Premise: Buy vs Rent TCO Guide (2026)
Compare GPU cloud vs on premise infrastructure. Evaluate 3-year TCO, facility power, colocation fees, and the 65% utilization breakeven threshold.
What Is GPU Cloud Computing? Architecture & Cost Guide
Learn what is GPU cloud computing, how hardware virtualization works, why GPUs beat CPUs for AI workloads, and how to choose between major cloud providers.
DeepSeek V4 Hardware Requirements: Pro & Flash Sizing
Calculate GPU cluster VRAM, node architecture, and cloud costs for DeepSeek V4 Pro (1.6T MoE) and V4 Flash (284B MoE) across H200 and B200 servers.
How we pick
GPU Picks is independently run by Ahmad Nugraha. We research public pricing, specs, and provider docs, and disclose affiliate relationships. Read our methodology →