How much does an eight-H100 node cost?
Multiply the published per-GPU rate by eight, then by the hours you plan to rent. Use the GPU-count filter on the H100 page to compare actual eight-GPU configurations; a single-GPU offer does not establish an eight-GPU node price.
| GPU | VRAM | $/hr | Where | |
|---|---|---|---|---|
| H100 (unspecified) | 80 GB | $0.94 | Prime Intellect spot / interruptible | Rent → |
| H100 (unspecified) | 80 GB | $1.18 | Latitude.sh reserved / committed | Rent → |
| H100 NVL | 94 GB | $1.29 | Microsoft Azure spot / interruptible | Rent → |
| H100 NVL | 94 GB | $1.29 | Microsoft Azure spot / interruptible | Rent → |
| H100 (unspecified) | 80 GB | $1.71 | Latitude.sh reserved / committed | Rent → |
| H100 (unspecified) | 80 GB | $1.75 | Hyperstack reserved / committed | Rent → |
| H100 SXM | 80 GB | $1.91 | TensorDock spot / interruptible | Rent → |
| H100 PCIe | 80 GB | $1.99 | RunPod community cloud | Rent → |
Check the total-per-hour column, GPU variant and region. CPU, RAM and interconnect can differ between nodes. The training calculator prices runtime at the selected node rate.
Use the job cost and fee table, model memory estimates, training calculator, and price trends for the relevant inputs.
Related GPUs
H100 rental prices · H100 vs A100 · H100 vs H200
Related questions
Numbers on this page come from today's verified snapshot. Full table on the homepage; method in the methodology.