Request access
Private Preview

H100-SXM5Hopper

NVIDIA H100 SXM5

Training and serving on 8-GPU HGX nodes, single node or multi-node over InfiniBand.

On-demand
$5.40 /GPU·hr
Spot
From $0.87 /GPU·hr
Reserved
Request a quote
Request this GPU

HBM3 per GPU

80 GB

NVIDIA reference figure

§01 Specifications Reviewed 2026-10-10
DWG 01HGX H100 8-GPU baseboard Source · NVIDIA DGX user guide
NVIDIA HGX H100 8-GPU baseboard, dimetric shop drawing
  1. 1. SXM5 module · 8 per baseboard, in a 2 × 4 grid
  2. 2. Heatsink · one per module; this one is drawn lifted
  3. 3. H100 GPU · one per module, under the heatsink
  4. 4. NVSwitch chip · 4 per baseboard, grouped at one end
  5. 5. Baseboard · joins the 8 GPUs in one NVLink domain

Point at a part or a row. The other one lights.

H100 SXM5 specifications.

TAB 01H100 SXM5 specifications Source · NVIDIA Reviewed 2026-10-10
H100 SXM5 specifications
Parameter Value Scope Source
Memory
Memory 80 GB HBM3 Per GPU [1]
Memory bandwidth 3.35 TB/s Per GPU [1]
Memory on one board 640 GB · 8 × 80 GB Per HGX board [1]
Compute, peak · dense / sparse
FP4 Tensor Core Not supported on Hopper Per GPU
FP8 Tensor Core 1,979 / 3,958 TFLOPS Per GPU [2]
BF16 Tensor Core 989 / 1,979 TFLOPS Per GPU [2]
Interconnect
NVLink 4 900 GB/s (18 links) Per GPU [3]
NVLink domain 8 GPUs per HGX H100 board Per HGX board [4]
Host interface PCIe Gen5 x16 · 128 GB/s Per GPU [1]
Power and form factor
Power Up to 700 W, configurable Per GPU [1]
Form factor SXM5 module on the HGX H100 baseboard [1]
Board 8 GPUs per HGX H100 board · 4 NVSwitch chips (third generation) Per HGX board [4]
  1. [1] NVIDIA H100 Tensor Core GPU
  2. [2] NVIDIA H100 datasheet
  3. [3] NVIDIA Hopper Architecture In-Depth
  4. [4] NVIDIA DGX H100/H200 User Guide
Reviewed 2026-10-10

Specifications are NVIDIA reference figures.

Best for

  • Training and fine-tuning
  • Inference
  • HPC (FP64 34 TFLOPS, FP64 Tensor 67 TFLOPS)

Notes on the figures

  • FP4: not supported on Hopper.
  • FP8 and BF16: NVIDIA datasheet footnote, "Shown with sparsity. Specifications 1/2 lower without sparsity."
§02 Price and terms USD per GPU-hour

On-demand, spot or reserved.

Metered by the second, billed hourly. Reserved rates are quoted per term.

  • On-demand

    $5.40 /GPU·hr

    Per GPU-second
    $0.001500
    Per 8 GPUs, per hour
    $43.20
    Per GPU-month, 730 h
    $3,942.00

    Preview list price. It applies when your account opens.

  • Spot

    From $0.87 /GPU·hr

    Work that checkpoints and can wait. Follow the current spot price, or set a maximum hourly price. If the spot price rises above your max, the workload may stop.

    "From" is the lowest spot price for that GPU. The spot price moves with supply and demand. The spot price always stays below the on-demand rate for the same GPU.

    Spot price moves with supply and demand. Follow it, or set a max price per GPU-hour: you pay the spot price, and nodes stop if it rises above your max. A max price limits what you pay. It does not reserve capacity. To guarantee capacity, reserve it. How spot works

  • Reserved

    Request a quote

    Capacity you need on a date, guaranteed. 1 month, 6 months or 12+ months.

    Reservations

Ways to pay Prepaid credit, pay-as-you-go and Enterprise. Each one covers on-demand, spot and reserved capacity. How billing works

EQ 01H100 SXM5 terms Source · Fantasti price list Reviewed 2026-10-10
Terms for NVIDIA B300, USD per GPU-hour. The on-demand rate is the preview list price of $10.45. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$10.45 /GPU·hr
1 monthTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
6 monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
12+ monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
Terms for NVIDIA B200, USD per GPU-hour. The on-demand rate is the preview list price of $9.35. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$9.35 /GPU·hr
1 monthTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
6 monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
12+ monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
Terms for NVIDIA H200 SXM, USD per GPU-hour. The on-demand rate is the preview list price of $5.94. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$5.94 /GPU·hr
1 monthTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
6 monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
12+ monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
Terms for NVIDIA H100 SXM5, USD per GPU-hour. The on-demand rate is the preview list price of $5.40. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$5.40 /GPU·hr
1 monthTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
6 monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
12+ monthsTake-or-pay · Held for the term, on one fabricTake-or-payHeld for the term, on one fabricRequest a quote
Terms for NVIDIA RTX PRO 6000 Blackwell Server Edition, USD per GPU-hour. The on-demand rate is the preview list price of $2.16. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$2.16 /GPU·hr
1 monthTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote
6 monthsTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote
12+ monthsTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote
Terms for NVIDIA L40S, USD per GPU-hour. The on-demand rate is the preview list price of $1.86. Reserved rates without a published figure are quoted.
TermCommitmentCapacityRate
On-demandNone · Not heldNoneNot held$1.86 /GPU·hr
1 monthTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote
6 monthsTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote
12+ monthsTake-or-pay · Held for the termTake-or-payHeld for the termRequest a quote

List price for the base shape: 1 GPU, 8 vCPU, 32 GiB. Larger shapes are quoted.

Reserved rates are quoted per term. Reservations are tied to a GPU type, a region and one InfiniBand fabric. Take-or-pay under a signed order form. No early cancellation without penalty (Terms §3).

§03 Node shape Platform configuration

Runs as instances and as cluster nodes.

HGX H100: 8 GPUs, 4 third-generation NVSwitch chips, 640 GB of HBM3 per node.

H100 machine shapes. Source · Fantasti platform configuration.

H100 runs as a 1-GPU instance (16 vCPU · 200 GiB) or as an 8-GPU node (128 vCPU · 1,600 GiB). 8-GPU nodes join one InfiniBand fabric at 400 Gb/s per GPU, 3.2 Tb/s per node. Host network adapter: 100 Gb/s per machine. Connections: 8 GPUs to One InfiniBand fabric, 8 links: 8 × 400 Gb/s; More 8-GPU nodes to One InfiniBand fabric; 1 GPU to Host network adapter.

  • 1 GPUInstance
    vCPU
    16
    RAM
    200 GiB
    1 GPU
  • Host network adapter
    rating
    100 Gb/s
    scope
    per machine
  • 8 GPUsNode
    HGX H100 8-GPU baseboard, line drawing
    vCPU
    128
    RAM
    1,600 GiB
    CPU
    Intel Xeon Platinum
    8468
    8 GPUs · NVLink 4
  • One InfiniBand fabricShared by the cluster
    per GPU
    400 Gb/s
    per node
    3.2 Tb/s

    Every node of one cluster sits on the same fabric. A cluster is placed onone fabric and stays there.

  • More 8-GPU nodesSame shape
    HGX H100 8-GPU baseboard, line drawing
    HGX H100 8-GPU baseboard, line drawing
    HGX H100 8-GPU baseboard, line drawing
  • 1 GPUInstance
    vCPU
    16
    RAM
    200 GiB
    1 GPU
  • Host network adapter
    rating
    100 Gb/s
    scope
    per machine
  • 8 GPUsNode
    HGX H100 8-GPU baseboard, line drawing
    vCPU
    128
    RAM
    1,600 GiB
    CPU
    Intel Xeon
    Platinum 8468
    8 GPUs · NVLink 4
  • One InfiniBand fabric
    per GPU
    400 Gb/s
    per node
    3.2 Tb/s
  • More 8-GPU nodesSame shape
  • 1 GPUInstance
    vCPU
    16
    RAM
    200 GiB
    1 GPU
  • Host network adapter
    rating
    100 Gb/s
    scope
    per machine
  • 8 GPUsNode
    HGX H100 8-GPU baseboard, line drawing
    vCPU
    128
    RAM
    1,600 GiB
    CPU
    Intel Xeon Platinum 8468
    8 GPUs · NVLink 4
  • One InfiniBand fabric
    per GPU
    400 Gb/s
    per node
    3.2 Tb/s
  • More 8-GPU nodesSame shape

H100 SXM5 listing.

CPU, RAM, local NVMe and the fabric vary by pool and are reported at placement. GPUs and InfiniBand adapters are passed through whole to one machine.

Status
Private Preview
Runs as
Single instances and GPU Clusters
Capacity
On-demand · Spot · Reserved
Metering
Metered by the second, billed hourly

GPU Clusters

TAB 02H100 SXM5 node shape Source · Fantasti platform configuration Reviewed 2026-10-10
H100 SXM5 node shape
ParameterValueScope
1 GPU16 vCPU · 200 GiBPer machine
8 GPUs128 vCPU · 1,600 GiBPer machine
Host CPUIntel Xeon Platinum 8468Per 8-GPU node
InfiniBand400 Gb/sPer GPU
InfiniBand3.2 Tb/sPer 8-GPU node
Host network adapter100 Gb/sPer machine

NVIDIA H100 SXM5 on Fantasti: fantasti.ai/gpus/h100-sxm5 · Request capacity: fantasti.ai/contact

Sheet
01 / 01
Title
H100 SXM5 datasheet
Figures
NVIDIA reference
Reviewed
2026-10-10
§04 Datasheets 8 listings
§05 Request Private Preview

Request H100 SXM5 capacity.

Tell us the count, the term and where it should run. We reply to a reviewed request, usually within one business day. A reserved term is quoted.

Request this GPU GPU catalog

Next