The NVIDIA H100 is still the workhorse of LLM training in 2026 — and renting one has never been cheaper. Prices fell sharply through 2025-2026 as supply caught up with demand. Here is where to get one cheapest today.
Cheapest H100 rentals, ranked (June 2026)
| Provider | Variant | On-demand /hr | Spot / community |
|---|---|---|---|
| Hyperstack | H100 PCIe | ~$1.90 | $1.52 (spot) |
| Vultr | H100 SXM | ~$2.30 | — |
| RunPod | H100 PCIe | $2.39 | $1.99 (Community) |
| Verda (ex-DataCrunch) | H100 SXM | $3.25 | spot lower |
| Lambda | H100 SXM | $3.29 | — |
| RunPod | H100 SXM | $3.29 | $1.99 (Community) |
| Crusoe | H100 SXM | $3.90 | — |
| Together AI | H100 SXM (cluster) | $4.79 | — |
| CoreWeave | HGX H100 SXM | $6.16 | — |
| AWS | H100 SXM (p5/8) | ~$6.88 | ~$3.37 (spot) |
| GCP | H100 SXM (a3/8) | ~$10.98 | ~$4.98 (spot) |
| Azure | H100 SXM (ND/8) | ~$12.29 | ~$2.27 (spot) |
Snapshot June 2026 — prices change weekly; verify on each provider’s pricing page. See the live cheapest H100 ranking and the full H100 SXM and H100 PCIe price tables.
The cheapest tier: community and interruptible
If your training job checkpoints and can survive a restart, the discounted tiers beat everything:
- RunPod Community Cloud — H100 from ~$1.99/hr (third-party hosts).
- Vast.ai interruptible — H100 SXM around $0.75-1.10/hr at the low end (a marketplace, host-set and variable).
The trade-off is interruption: your instance can be reclaimed. For checkpointed training and batch jobs that is usually fine. See on-demand vs spot.
Why hyperscalers cost 3-5x more
AWS, GCP and Azure sell flagship H100s only as whole 8-GPU instances (p5, a3, ND). The instance bundles dozens of vCPUs, hundreds of GB of RAM and enterprise support, so the per-GPU figure (instance price / 8) lands at $7-12/hr. You pay that premium for the ecosystem, regions and reliability — not for the silicon. For most independent ML work, a neocloud is far cheaper.
How to pick
- Need NVLink / multi-GPU? Choose SXM (Vultr, Lambda, RunPod, Verda).
- Single-GPU job, tightest budget? H100 PCIe on Hyperstack (~$1.90) or RunPod.
- Checkpointed / batch? RunPod Community or Vast.ai interruptible.
- Need a full node for big distributed training? See the cheapest 8xH100 node.
Price your exact job — GPUs x hours across every provider — in the cost calculator.