NVIDIA H200

Rent H200 GPUs on demand for LLM training and inference with 141 GB of memory per GPU. Choose a template, GPU VM or cluster for your workload.

Launch H200Compare GPU pricing

From $3.99/GPU/hour · USD on-demand rate · Storage extra

H200 rental

Choose how you rent H200.

Start on demand, use interruptible capacity for restartable jobs, or plan a longer reservation with our team.

USD rates per GPU-hour. Storage is billed separately; regional prices and availability vary.

  • On demand
    GPU rate
    $3.99/hr
    When to choose it
    Pay by the minute without a long-term commitment. Choose your GPU count and region in the dashboard.
  • Spot
    GPU rate
    $1.99/hr
    When to choose it
    Lower-cost, interruptible capacity. Save checkpoints and use jobs that can restart; subject to availability.
  • Reserved capacity
    GPU rate
    Discuss your requirements
    When to choose it
    Plan GPU count, reservation term and networking with our team for sustained or multi-node workloads.

Compare the full cost

Review H200 rental options, ownership costs and the tradeoffs against H100 before choosing your setup.

Read the H200 pricing guide

Plan multi-GPU inference

Understand tensor, pipeline and data parallelism before distributing a model across GPUs.

Read the inference scaling guide

Reserve a cluster

Share your workload and schedule to confirm the right configuration and capacity.

Discuss reserved H200 capacity

Workloads

What to run on H200.

Choose for your workload’s memory needs, software support and measured runtime.

Larger model working sets

Keep more weights, activations and inference cache on each GPU. Size your job for its precision, context length and batch size.

Training and fine-tuning

Use the additional memory for larger batches or longer sequences when your training framework supports them.

Distributed workloads

Use an H200 cluster when the job needs multiple nodes. Plan model parallelism and data movement alongside GPU count.

Choose your setup

Start with the environment you need.

Use a preconfigured container or manage your operating system in a GPU VM. The dashboard shows current GPU and region availability.

GPU Templates

Start with PyTorch, ComfyUI or another supported environment. Drivers and framework dependencies are preconfigured.

Explore templates

GPU VMs

Get SSH and full root access when you need control over the operating system and runtime. Available configurations vary by GPU and region.

Explore VMs

GPU clusters

Instant and reserved H200 clusters are available. Check the dashboard for current instant capacity.

Explore clusters

Sizing

Start with memory. Then measure performance.

Model weights are only part of the working set. Leave room for everything the job needs while it runs.

For inference

Account for model precision, context length, KV cache and concurrent requests. A model loading successfully does not tell you how much traffic it can serve.

For training

Include activations, gradients and optimizer state. Batch size, sequence length and checkpointing change memory use.

For multiple GPUs

GPU memory is not automatically pooled. Use a framework and parallelism strategy that distribute the workload across devices.

Hardware specifications: NVIDIA H200. Software and workload affect realized performance.

Compare options

Compare memory and hourly rates.

Use this as a shortlist, then test your workload. Lower hourly pricing does not always mean a lower total job cost.

Published USD on-demand rates per GPU; storage extra. Availability and regional prices vary.

  • NVIDIA H200
    Memory
    141 GB HBM3e
    Architecture
    Hopper
    From / GPU / hour
    $3.99
  • NVIDIA H100
    Memory
    80 GB HBM3
    Architecture
    Hopper
    From / GPU / hour
    $2.69
  • NVIDIA B200
    Memory
    180 GB per GPU¹
    Architecture
    Blackwell
    From / GPU / hour
    Request a quote
  • NVIDIA RTX PRO 6000
    Memory
    96 GB GDDR7
    Architecture
    Blackwell
    From / GPU / hour
    $1.89
  • NVIDIA A100 80 GB
    Memory
    80 GB HBM2e
    Architecture
    Ampere
    From / GPU / hour
    $1.49
  • NVIDIA A100 40 GB
    Memory
    40 GB HBM2
    Architecture
    Ampere
    From / GPU / hour
    $0.89
  • NVIDIA L4
    Memory
    24 GB GDDR6
    Architecture
    Ada Lovelace
    From / GPU / hour
    $0.44

Same model, same serving stack, both GPUs: Read the measured H100 vs H200 comparison

Before you start

Common questions.

Practical details for choosing and using this product.

How do I get started with NVIDIA H200?

Choose a template or VM, select the GPU and region, and review the configuration and price in the dashboard before launch.

Is storage included in the GPU rate?

Storage is billed separately. Retained storage continues to incur charges while an instance is paused. Review the full configuration price before launching.

Will my model fit on one GPU?

This GPU has 141 GB HBM3e. Fit depends on weights, precision, framework overhead and workload state. For inference, also account for context and concurrency; for training, include activations and optimizer state.

Can I get more than eight GPUs?

For workloads spanning nodes, explore GPU clusters. Reserved capacity can be planned at 128, 256, 1,024 GPUs and beyond, subject to configuration and availability.

Get started with NVIDIA H200.

Launch a GPU instance or talk to our team about the right setup for your workload.

Launch H200