The cheapest H100 PCIe on the table right now is $0.347 per GPU-hour at Vast.ai (spot), out of 21 offers from 10 providers. The H100 PCIe carries 80 GB of VRAM on NVIDIA's Hopper generation. That is peer/marketplace capacity — rented from other users' machines, and not directly comparable to a dedicated instance.
On-demand pricing for the same card spans $1.33 to $3.61 per GPU-hour — a 2.7× spread between the cheapest and the most expensive provider for identical silicon. That gap is the entire reason this table exists.
Cheapest by pricing model: on-demand $1.33 at Vast.ai · spot $0.347 at Vast.ai.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| Vast.ai | marketplace | $0.347 | spot | 2× GPU | — | yes | rent |
| Vast.ai | marketplace | $1.20 | spot | 1, 3× GPU | — | yes | rent |
| Vast.ai | marketplace | $1.33 | spot | 8× GPU | — | yes | rent |
| Vast.ai | marketplace | $1.33 | on-demand | 1–3× GPU | — | yes | rent |
| Lium.io | marketplace | $1.50 | on-demand | 1× GPU | — | yes | rent |
| Vast.ai | marketplace | $1.60 | spot | 4× GPU | — | yes | rent |
| Prime Intellect | dedicated | $2.35 | on-demand | 1× GPU | — | yes | rent |
| Hyperstack | dedicated | $2.60 | on-demand | 1× GPU | — | — | rent |
| Runpod | dedicated | $2.89 | on-demand | 1× GPU | — | yes | rent |
| Runpod | dedicated | $2.89 | spot | 1× GPU | — | yes | rent |
| Vast.ai | marketplace | $3.07 | on-demand | 4× GPU | — | yes | rent |
| Thunder Compute | dedicated | $3.20 | on-demand | 1× GPU | — | yes | rent |
| Lambda Labs | dedicated | $3.29 | on-demand | 1× GPU | — | — | rent |
| Vast.ai | marketplace | $3.33 | on-demand | 8× GPU | — | yes | rent |
| Scaleway | dedicated | $3.34 | on-demand | 1, 2× GPU | — | — | rent |
| Beam | dedicated | $3.55 | on-demand | 1× GPU | — | yes | rent |
| OVHcloud | dedicated | $3.61 | on-demand | 1× GPU | — | — | rent |
How much does an H100 PCIe cost per hour?
As of the latest scrape, $0.347 per GPU-hour at Vast.ai is the cheapest H100 PCIe offer across 10 tracked cloud providers. The cheapest on-demand rate is $1.33 per GPU-hour at Vast.ai, and the most expensive on-demand rate is $3.61 at OVHcloud.
Which cloud is cheapest for the H100 PCIe?
Vast.ai at $0.347 per GPU-hour on spot capacity. Prices move constantly, so this page is regenerated from the live scrape every 15 minutes. Note that marketplace tiers are peer hardware and are not directly comparable to dedicated capacity.
Is spot H100 PCIe capacity cheaper?
Yes — the cheapest spot H100 PCIe is $0.347 per GPU-hour at Vast.ai, -74% versus the cheapest on-demand rate of $1.33. Spot instances can be reclaimed by the provider at any time, so they suit checkpointed training and batch inference rather than long-lived services.
How many GB of VRAM does the H100 PCIe have?
80 GB, on the Hopper architecture. Multi-GPU instances multiply that: an 8× H100 PCIe node exposes 640 GB of GPU memory in total.