Real-time NVIDIA H100 SXM5 cloud pricing from 28 providers. Cheapest on-demand: $1.79/hr (Shadeform). Updated daily by GridStackHub.
Sorted by cheapest per-GPU hourly rate. Includes on-demand, spot, and reserved pricing where available.
| Provider | Price/hr | Type | Region | VRAM | Updated |
|---|---|---|---|---|---|
|
RunPodLowest
NVIDIA H100 PCIe-spot
|
$1.35/hr | Spot | community | 80 GB | Tue Apr 21 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
NVIDIA H100 NVL-spot
|
$1.40/hr | Spot | community | 94 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Vast.ai
H100 SXM (marketplace)
|
$1.49/hr | Spot | Various | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
NVIDIA H100 80GB HBM3-spot
|
$1.50/hr | Spot | community | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
DataCrunch
H100 SXM5 (Spot)
|
$1.65/hr | Spot | EU (Finland) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
CoreWeave
H100 SXM5 Reserved
|
$1.79/hr | Reserved 1yr | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Shadeform
H100 SXM (best price)
|
$1.79/hr | On-Demand | Various | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
H100 SXM (Community)
|
$1.89/hr | Spot | US/EU | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
NVIDIA H100 PCIe
|
$1.99/hr | On-Demand | us-east-1 | 80 GB | Tue Apr 21 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lambda
1x H100 SXM
|
$1.99/hr | On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Together AI
H100 (Reserved Instances)
|
$1.99/hr | On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lambda
8x H100 SXM
|
$1.99/hr $15.92/hr for 8× node |
On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Crusoe Energy
H100 SXM (Climate-Aligned)
|
$2.06/hr | On-Demand | US (Texas) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
TensorDock
H100 SXM 80GB
|
$2.09/hr | On-Demand | US/EU | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
FluidStack
H100 SXM5 80GB
|
$2.15/hr | On-Demand | US/EU | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Crusoe Cloud
h100-80gb-sxm-ib-1x
|
$2.17/hr | On-Demand | us-central | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Crusoe Cloud
h100-80gb-sxm-ib-8x
|
$2.17/hr $17.36/hr for 8× node |
On-Demand | us-central | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Oblivus Cloud
H100 SXM5
|
$2.19/hr | On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
DataCrunch
H100 SXM5 80GB
|
$2.20/hr | On-Demand | EU (Finland) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
CoreWeave
H100 SXM5
|
$2.23/hr | On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Genesis Cloud
H100 SXM5 80GB
|
$2.35/hr | On-Demand | EU (Iceland) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
FluidStack
h100_80gb_sxm5_x1
|
$2.40/hr | On-Demand | us | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Nebius
H100 SXM5 (gpu-h100-b)
|
$2.40/hr | On-Demand | EU (Finland) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
FluidStack
h100_80gb_sxm5_x8
|
$2.40/hr $19.20/hr for 8× node |
On-Demand | us | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
AWS
p5.48xlarge (8x H100)
|
$2.40/hr $19.22/hr for 8× node |
Reserved 1yr | us-east-1 | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Google Cloud
a3-highgpu-8g (8x H100)
|
$2.45/hr $19.63/hr for 8× node |
Reserved 1yr | us-central1 | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lambda Labs
gpu_1x_h100_sxm5_80gb
|
$2.49/hr | On-Demand | us-east-1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
H100 SXM 80GB
|
$2.49/hr | On-Demand | US/EU | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Hetzner
GX11 (1x H100 SXM5)
|
$2.49/hr | On-Demand | EU (Germany) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lambda Labs
gpu_8x_h100_sxm5_80gb
|
$2.49/hr $19.92/hr for 8× node |
On-Demand | us-east-1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
TensorDock
tensordock-h100
|
$2.50/hr | Spot | global | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Scaleway
H100-1-80G
|
$2.55/hr | On-Demand | EU (France) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Azure
ND H100 v5 (8x H100)
|
$2.56/hr $20.49/hr for 8× node |
Reserved 1yr | East US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
NVIDIA H100 NVL
|
$2.59/hr | On-Demand | us-east-1 | 94 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
RunPod
NVIDIA H100 80GB HBM3
|
$2.69/hr | On-Demand | us-east-1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Latitude.sh
GPU.H100x1
|
$2.79/hr | On-Demand | US/BR | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
OVHcloud
GPU H100 SXM5
|
$2.80/hr | On-Demand | EU (France) | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Cirrascale
H100 SXM5 (Dedicated)
|
$2.80/hr $22.40/hr for 8× node |
On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Paperspace
H100 80GB SXM
|
$3.09/hr | On-Demand | US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Oracle Cloud
BM.GPU.H100.8 (8x H100)
|
$3.21/hr $25.68/hr for 8× node |
On-Demand | us-ashburn-1 | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lepton AI
h100-sxm5-8-reserved
|
$3.40/hr $27.20/hr for 8× node |
Reserved 1yr | us-east | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Jarvis Labs
H100_x1
|
$3.49/hr | On-Demand | us | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Jarvis Labs
H100_x8
|
$3.49/hr $27.92/hr for 8× node |
On-Demand | us | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Genesis Cloud
gpu-h100-x1
|
$3.59/hr | On-Demand | eu-west | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Genesis Cloud
gpu-h100-x8
|
$3.59/hr $28.72/hr for 8× node |
On-Demand | eu-west | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Paperspace
H100-80G
|
$3.89/hr | On-Demand | us-east | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Paperspace
H100x8
|
$3.89/hr $31.12/hr for 8× node |
On-Demand | us-east | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Google Cloud
a3-highgpu-8g (8x H100)
|
$3.90/hr $31.21/hr for 8× node |
On-Demand | us-central1 | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
CoreWeave
H100_80GB_SXM5_x1
|
$3.92/hr | On-Demand | US | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
CoreWeave
H100_80GB_SXM5_x4
|
$3.92/hr $15.68/hr for 4× node |
On-Demand | US | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
CoreWeave
H100_80GB_SXM5_x8
|
$3.92/hr $31.36/hr for 8× node |
On-Demand | US | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lepton AI
h100-sxm5-1
|
$4.00/hr | On-Demand | us-east | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
DataCrunch
1xH100-80G-SXM5
|
$4.00/hr | On-Demand | fi-01 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
DataCrunch
8xH100-80G-SXM5
|
$4.00/hr $32.00/hr for 8× node |
On-Demand | fi-01 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Lepton AI
h100-sxm5-8
|
$4.00/hr $32.00/hr for 8× node |
On-Demand | us-east | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
AWS
p5.48xlarge (8x H100)
|
$4.10/hr $32.77/hr for 8× node |
On-Demand | us-east-1 | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
GCP
a3-highgpu-8g
|
$4.10/hr $32.77/hr for 8× node |
On-Demand | us-central1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Azure
ND H100 v5 (8x H100)
|
$4.10/hr $32.78/hr for 8× node |
On-Demand | East US | 80 GB | Sun Apr 12 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
Latitude.sh
c3.large.x86-gpu-h100-80gb-8
|
$4.38/hr $35.00/hr for 8× node |
On-Demand | us | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
OVHcloud
h100-1
|
$4.50/hr | On-Demand | eu-west | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
OVHcloud
h100-8
|
$4.50/hr $36.00/hr for 8× node |
On-Demand | eu-west | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
AWS
p5.48xlarge
|
$12.29/hr $98.32/hr for 8× node |
On-Demand | us-east-1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
GCP
a3-megagpu-8g
|
$12.29/hr $98.32/hr for 8× node |
On-Demand | us-central1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
|
AWS
p5e.48xlarge
|
$13.50/hr $108.00/hr for 8× node |
On-Demand | us-east-1 | 80 GB | Thu Apr 23 2026 00:00:00 GMT+0000 (Coordinated Universal Time) |
Key hardware specifications for the NVIDIA H100 SXM5.
| Architecture | Hopper (SXM5) |
| VRAM | 80GB HBM3 |
| Memory Bandwidth | 3.35 TB/s |
| FP8 Throughput | 3,958 TFLOPS |
| FP16 Throughput | 1,979 TFLOPS |
| GPU-to-GPU Bandwidth | 900 GB/s (NVSwitch) |
| TDP | 700W |
| Gen | 4th Gen NVLink |
The NVIDIA H100 SXM5 remains the workhorse of AI infrastructure in 2026. With 80GB HBM3 memory, 3.35 TB/s memory bandwidth, and 3,958 TFLOPS FP8 compute, it handles the full range of production workloads — from fine-tuning 70B parameter models to multi-node distributed training.
H100 SXM5 on-demand pricing has dropped significantly as supply expanded. Sub-$2/hr from Lambda and CoreWeave makes it the best-value high-performance GPU for most teams. Vast.ai spot pricing can push below $1.50/hr for interruptible workloads.
The SXM5 form factor matters for multi-GPU jobs: NVSwitch interconnect provides 900 GB/s GPU-to-GPU bandwidth per GPU, enabling tight all-to-all coupling across 8-GPU nodes. For single-GPU inference, PCIe H100 is cheaper. For 64-GPU+ training clusters, SXM5 is the standard.
Reserved pricing (1-year) saves 20–40% at CoreWeave and Google Cloud. For teams running H100 continuously, reserved contracts typically break even versus on-demand within 60–90 days.
Use GridStackHub's GPU cost calculator to get a ranked comparison with hidden-cost breakdown (egress + storage) across all providers.
📊 Open Calculator View All GPU Pricing