Back to all GPUs

NVIDIA B200 Cloud GPU Specs

HexGrid Cloud rents the NVIDIA B200 at $7.50 per GPU hour, billed per minute. The B200 carries 180GB of HBM3e memory on the Blackwell architecture.

180GB VRAMBlackwellDeploys in ~1 minSecure Cloud

HexGrid Cloud price

$7.50/hr

Billed per minute

VRAM

180GB

HBM3e

FP32

Compute performance

Market average

$6.79/hr

Other clouds, list price

Running the B200 on HexGrid Cloud

What you get when you deploy with us, beyond the hourly rate.

Live in under a minute

Pick a region, hit deploy, and SSH in. No quota requests, no sales call.

Per-minute billing

You pay for the minutes you use. Stop the instance and billing stops with it.

Ready for training day one

CUDA, PyTorch and the usual drivers are preinstalled, or bring your own image.

No egress charges

Move checkpoints and datasets out without a surprise bandwidth bill.

B200 pricing across the market

Published on-demand rates from other clouds, so you can see where our B200 price sits.

Market average $6.79/hr across 1 cloud

  • HexGrid CloudYou're here$7.50/hr
  • RunPod$6.79/hr

Other providers' rates are their published on-demand list prices. They exclude committed-use discounts and can change at any time. Shown for reference only.

About the NVIDIA B200

What the B200 is built for, and where it falls short.

The NVIDIA B200 is a Blackwell accelerator designed for modern AI factories. Each GPU provides 180GB of HBM3e memory with up to 8 TB/s of memory bandwidth and introduces second-generation Transformer Engine technology with FP4 acceleration for large language models, generative AI, and reasoning workloads.

Best suited for

The B200 combines Blackwell Tensor Cores, 180GB of HBM3e, up to 8 TB/s memory bandwidth, FP4 acceleration, and fifth-generation NVLink for extremely large AI workloads.

  • LLM training
  • LLM inference
  • AI reasoning
  • Mixture-of-Experts models
  • Generative AI
  • HPC

Strengths

  • 180GB HBM3e memory
  • Up to 8 TB/s memory bandwidth
  • FP4 acceleration
  • 1.8 TB/s NVLink
  • MIG and enterprise virtualization support

Limitations

  • Very high power requirements
  • Premium cloud pricing
  • Requires specialized datacenter infrastructure

NVIDIA B200 specifications

Full technical specifications for the NVIDIA B200.

VendorNVIDIAArchitectureBlackwellCategoryDatacenter GPUReleased2024Form factorSXMInterfaceSXM

Memory

Memory
180 GB
Type
HBM3e
Bandwidth
8,000 GB/s
ECC
Yes

Compute

Tensor generation
Gen 5

Power & physical

TDP
1000 W
Form factor
SXM
Cooling
Server-cooled

Interconnect

Interface
SXM
NVLink
Yes
NVLink bandwidth
1,800 GB/s
Infinity Fabric
No

Virtualization

MIG
Yes
vGPU
Yes
SR-IOV
Yes

Software

CUDA
Yes
ROCm
No

Precision support

FP64
Yes
FP32
Yes
TF32
Yes
BF16
Yes
FP16
Yes
FP8
Yes
FP6
Yes
FP4
Yes
INT8
Yes

B200 frequently asked questions

Common questions about NVIDIA B200 pricing, specifications and workloads on HexGrid Cloud.

How much memory does NVIDIA B200 have?

The NVIDIA B200 provides 180GB of HBM3e memory per GPU.

What is the B200 memory bandwidth?

NVIDIA HGX B200 systems provide up to 8 TB/s of HBM3e memory bandwidth per GPU.

Does B200 support FP4?

Yes. Blackwell Tensor Cores and the second-generation Transformer Engine support FP4 AI computation.

Does B200 support MIG?

Yes. NVIDIA provides MIG-backed virtual GPU configurations for HGX B200.

How much does a B200 cost per hour?

B200 cloud pricing varies significantly by provider and cluster configuration. HexGrid compares available hourly B200 offers.

Other GPUs on HexGrid Cloud

If the B200 isn't the right fit, these are also available to deploy.

Browse every GPU we rent

Start training on the B200 today

$7.50 per GPU hour, billed per minute, running in about a minute.

Specification sources