The cheapest H100 SXM on the table right now is $0.133 per GPU-hour at Vast.ai (spot), out of 68 offers from 33 providers. The H100 SXM 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 $1.34 to $24.00 per GPU-hour — an 18.0× 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 $1.34 at Vast.ai · spot $0.133 at Vast.ai · reserved $1.89 at fal.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| Vast.ai | marketplace | $0.133 | spot | 1× GPU | — | yes | rent |
| Vast.ai | marketplace | $0.447 | spot | 2, 4× GPU | — | yes | rent |
| Google Cloud | dedicated | $0.970 | spot | 1× GPU | — | — | rent |
| Vast.ai | marketplace | $1.34 | on-demand | 1× GPU | — | yes | rent |
| Verda | dedicated | $1.63 | spot | 1–8× GPU | — | — | rent |
| Novita | dedicated | $1.70 | spot | 1× GPU | — | — | rent |
| Vast.ai | marketplace | $1.73 | on-demand | 2× GPU | — | yes | rent |
| Vast.ai | marketplace | $1.78 | on-demand | 4× GPU | — | yes | rent |
| Cudo Compute | dedicated | $1.79 | on-demand | 1× GPU | — | yes | rent |
| GPU.ai | dedicated | $1.87 | on-demand | 1× GPU | — | yes | rent |
| fal | dedicated | $1.89 | reserved | 1× GPU | — | — | rent |
| AWS | dedicated | $1.92 | spot | 8× GPU | — | — | rent |
| Voltage Park | dedicated | $1.99 | on-demand | 1× GPU | — | yes | rent |
| Azure | dedicated | $2.04 | spot | 8× GPU | — | — | rent |
| Nebius | dedicated | $2.15 | reserved | 8× GPU | — | — | rent |
| Seeweb | dedicated | $2.20 | on-demand | 1× GPU | — | — | rent |
| LeaderGPU | dedicated | $2.37 | reserved | 8× GPU | 1 mo | — | rent |
| Denvr Dataworks | dedicated | $2.45 | on-demand | 8× GPU | — | — | rent |
| AWS | dedicated | $2.48 | spot | 1× GPU | — | — | rent |
| Hyperstack | dedicated | $2.50 | on-demand | 1× GPU | — | — | rent |
| Spheron | dedicated | $2.64 | on-demand | 1× GPU | — | — | rent |
| Jarvis Labs | dedicated | $2.69 | on-demand | 1× GPU | — | yes | rent |
| Massed Compute | dedicated | $2.73 | on-demand | 1–4× GPU | — | — | rent |
| Lium.io | marketplace | $2.75 | on-demand | 8× GPU | — | yes | rent |
| Massed Compute | dedicated | $2.92 | on-demand | 8× GPU | — | — | rent |
| Civo | dedicated | $2.99 | on-demand | 1, 8× GPU | — | — | rent |
| Verda | dedicated | $3.25 | on-demand | 1–8× GPU | — | — | rent |
| DigitalOcean | dedicated | $3.26 | reserved | 8× GPU | 12 mo | — | rent |
| Runpod | dedicated | $3.29 | on-demand | 1× GPU | — | yes | rent |
| Runpod | dedicated | $3.29 | spot | 1× GPU | — | yes | rent |
| Novita | dedicated | $3.39 | on-demand | 1× GPU | — | — | rent |
| Scaleway | dedicated | $3.69 | on-demand | 8× GPU | — | — | rent |
| Together AI | dedicated | $3.69 | reserved | 8× GPU | — | — | rent |
| Scaleway | dedicated | $3.72 | on-demand | 4× GPU | — | — | rent |
| Nebius | dedicated | $3.85 | on-demand | 8× GPU | — | — | rent |
| Scaleway | dedicated | $3.85 | on-demand | 2× GPU | — | — | rent |
| Crusoe | dedicated | $3.90 | on-demand | 8× GPU | — | — | rent |
| Modal | dedicated | $3.95 | on-demand | 1× GPU | — | yes | rent |
| Lambda Labs | dedicated | $3.99 | on-demand | 8× GPU | — | — | rent |
| Together AI | dedicated | $3.99 | on-demand | 8× GPU | — | — | rent |
| Lambda Labs | dedicated | $4.09 | on-demand | 4× GPU | — | — | rent |
| Lambda Labs | dedicated | $4.19 | on-demand | 2× GPU | — | — | rent |
| Lambda Labs | dedicated | $4.29 | on-demand | 1× GPU | — | — | rent |
| Prime Intellect | dedicated | $4.29 | on-demand | 1× GPU | — | yes | rent |
| DigitalOcean | dedicated | $4.41 | on-demand | 8× GPU | — | — | rent |
| fal | dedicated | $4.50 | on-demand | 1× GPU | — | — | rent |
| Replicate | dedicated | $5.49 | on-demand | 1× GPU | — | yes | rent |
| Replicate | dedicated | $5.49 | reserved | 2–8× GPU | — | yes | rent |
| CoreWeave | dedicated | $6.16 | on-demand | 8× GPU | — | — | rent |
| Baseten | dedicated | $6.50 | on-demand | 1× GPU | — | yes | rent |
| AWS | dedicated | $6.88 | on-demand | 1, 8× GPU | — | — | rent |
| Google Cloud | dedicated | $9.80 | on-demand | 1× GPU | — | — | rent |
| Oracle Cloud | dedicated | $10.75 | on-demand | 8× GPU | — | — | rent |
| Azure | dedicated | $11.06 | on-demand | 8× GPU | — | — | rent |
| Latitude.sh | dedicated | $24.00 | on-demand | 1× GPU | — | — | rent |
How much does an H100 SXM cost per hour?
As of the latest scrape, $0.133 per GPU-hour at Vast.ai is the cheapest H100 SXM offer across 33 tracked cloud providers. The cheapest on-demand rate is $1.34 per GPU-hour at Vast.ai, and the most expensive on-demand rate is $24.00 at Latitude.sh.
Which cloud is cheapest for the H100 SXM?
Vast.ai at $0.133 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 SXM capacity cheaper?
Yes — the cheapest spot H100 SXM is $0.133 per GPU-hour at Vast.ai, -90% versus the cheapest on-demand rate of $1.34. 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 SXM have?
80 GB, on the Hopper architecture. Multi-GPU instances multiply that: an 8× H100 SXM node exposes 640 GB of GPU memory in total.