NVIDIA H200 rental price per hour
141GB HBM3e VRAM · Hopper architecture · released 2024
The cheapest way to rent an NVIDIA H200 by the hour in June 2026 is about $3.44/hr on Paperspace (DigitalOcean). We track 12 providers publishing an on-demand price for it , and spot/community rates reach as low as $2.38/hr. Memory-bound LLM inference and training of large models. Same compute as H100 but 141GB VRAM and far higher memory bandwidth — fits bigger models per GPU.
Source: Provider pricing pages. Data as of June 2026.
NVIDIA H200 price by provider
| Provider | Type | On-demand /hr | Spot/community /hr | Notes |
|---|---|---|---|---|
| Paperspace (DigitalOcean) | specialist | $3.44/hr | — | DigitalOcean GPU Droplet (HGX, per GPU). |
| Hyperstack | specialist | $3.50/hr | — | H200 SXM on-demand. |
| RunPod | specialist | $3.59/hr | — | Secure Cloud on-demand (low end of $3.59-$4.39 range). |
| Vast.ai | marketplace | $4.00/hr | — | Marketplace median. Host-set. |
| Verda (formerly DataCrunch) | specialist | $4.00/hr | — | SXM5 on-demand. |
| Crusoe | specialist | $4.29/hr | — | On-demand. |
| Nebius | specialist | $4.50/hr | $2.45/hr | HGX per GPU; reserved $2.45. |
| Together AI | specialist | $5.99/hr | — | GPU Cluster per-GPU; reserved from $3.99. |
| CoreWeave | specialist | $6.30/hr | — | HGX H200, per GPU. |
| AWS EC2 | hyperscaler | $7.91/hr | $2.38/hr | Derived: p5en.48xlarge $63.30/hr / 8 GPUs. |
| Microsoft Azure | hyperscaler | $10.60/hr | — | Derived: ND H200 v5 $84.80/hr / 8 GPUs (East US 2). |
| Google Cloud | hyperscaler | $10.85/hr | $5.35/hr | Derived: a3-ultragpu-8g $86.76/hr / 8 GPUs. |
| Lambda | specialist | — | — | On-demand not offered; H200 only via custom-quote 1-Click Clusters. |
| Vultr Cloud GPU | specialist | — | — | Not on Vultr's own price table (third-party cited 'from $2.20', unconfirmed). |
Source: Provider pricing pages. Data as of June 2026.
Sorted cheapest on-demand first; "—" means the provider does not publish a price for this GPU (see the note). Snapshot June 2026 — cloud GPU prices change weekly; verify on the provider's pricing page before relying on a figure.
About the NVIDIA H200
Memory-bound LLM inference and training of large models. Same compute as H100 but 141GB VRAM and far higher memory bandwidth — fits bigger models per GPU.
- VRAM: 141GB HBM3e
- Architecture: Hopper
- Released: 2024
- Cheapest on-demand: $3.44/hr on Paperspace (DigitalOcean)
- Cheapest spot/community: $2.38/hr on AWS EC2
Frequently asked questions
How much does it cost to rent an NVIDIA H200 per hour?
The cheapest on-demand NVIDIA H200 we track is $3.44/hr on Paperspace (DigitalOcean), with 12 providers publishing a price. Spot/community rates go as low as $2.38/hr. Snapshot June 2026 — cloud GPU prices change weekly; verify on the provider's pricing page before relying on a figure.
Which provider is cheapest for the NVIDIA H200?
Paperspace (DigitalOcean) at $3.44/hr is the cheapest on-demand option in our June 2026 snapshot. Specialist neoclouds and marketplaces usually beat the hyperscalers (AWS/GCP/Azure) by 2-4x on the same GPU.
What is the NVIDIA H200 good for?
Memory-bound LLM inference and training of large models. Same compute as H100 but 141GB VRAM and far higher memory bandwidth — fits bigger models per GPU. It has 141GB HBM3e of VRAM and uses the Hopper architecture (released 2024).
Is spot or on-demand cheaper for the NVIDIA H200?
Spot, community and interruptible tiers can cut the NVIDIA H200 price roughly in half — the cheapest discounted rate we track is $2.38/hr versus $3.44/hr on-demand. The trade-off is your instance can be reclaimed, so use spot for fault-tolerant or checkpointed jobs.
Related GPUs
- AMD Instinct MI300X (192GB HBM3)
- NVIDIA B200 (192GB HBM3e)
- NVIDIA H100 PCIe (80GB HBM2e)
- NVIDIA H100 SXM (80GB HBM3)
Keep exploring
Source & accuracy
Figures are a dated snapshot from each provider's published pricing page (as of June 2026). Hyperscaler per-GPU prices are derived from 8-GPU instance prices (instance price / 8) and include bundled vCPU/RAM; marketplace (Vast.ai) figures are moving medians, not list prices. We do not invent numbers — where a value is unknown we show "—" with a reason. See our methodology. This is an informational comparison, not a quote.
Last updated: 2026-06-21