H100 SXM pricing on Modal
Every Modal H100 SXM offering we track, normalized to a per-GPU hourly rate and ranked against 36 providers with current H100 SXM pricing.
The cheapest H100 SXM we currently track is $1.14/hr on Verda.
Modal H100 SXM rates
Hourly prices per GPU. Multi-GPU rows show the per-GPU rate, not the instance total.
| Price / GPU-hr | Type | GPUs |
|---|---|---|
| $3.95 | On-demand | 1 |
H100 SXM specifications
- VRAM
- 80GB
- Architecture
- Hopper
- CUDA Cores
- 14,592
- Tensor Cores
- 456
- Memory Bandwidth
- 3350 GB/s
- TDP
- 700W
H100 SXM on other providers
- Verda$1.14/hr
- Latitude.sh$1.66/hr
- Cudo Compute$1.79/hr
- fal.ai$1.89/hr
- Voltage Park$1.99/hr
- GMI Cloud$2.00/hr
- Nebius$2.15/hr
- Seeweb$2.18/hr
- Thunder Compute$2.19/hr
- 1Legion$2.20/hr
- Deep Infra$2.20/hr
- Zettabyte$2.20/hr
- Denvr Dataworks$2.30/hr
- Massed Compute$2.45/hr
- CoreWeave$2.46/hr
- Civo$2.49/hr
- Koyeb$2.50/hr
- IO.NET$2.60/hr
- Jarvis Labs$2.69/hr
- Runpod$2.69/hr
- Sesterce$2.75/hr
- Atlas Cloud$2.95/hr
- Hyperbolic$3.19/hr
- Hyperstack$3.20/hr
- Scaleway$3.31/hr
- Oblivus$3.50/hr
- AceCloud$3.63/hr
- Crusoe$3.90/hr
- Lambda Labs$3.99/hr
- DigitalOcean$4.41/hr
- Together AI$5.40/hr
- Paperspace$5.95/hr
- Amazon AWS$6.88/hr
- Oracle Cloud$10.00/hr
- Microsoft Azure$11.06/hr
About Modal
Modal is a serverless GPU platform that lets developers run Python functions, jobs and inference endpoints on NVIDIA GPUs with per-second billing and scale-to-zero.
Frequently Asked Questions
How much does the H100 SXM cost on Modal?
Modal lists the H100 SXM from $3.95 per GPU-hour across 1 tracked configuration. Prices are collected daily — see the table above for the current rates.
Is Modal the cheapest place to rent the H100 SXM?
Modal ranks #29 of 36 providers we track with current H100 SXM pricing. The cheapest right now is Verda. Price is only one factor — region availability, network, and storage costs differ between providers.
Which H100 SXM configurations does Modal offer?
We currently track 1 H100 SXM offering from Modal, covering the GPU counts, pricing types, and regions listed in the table above. Every price is normalized to a per-GPU hourly rate so it can be compared across providers.