Compute Comparison
NVIDIABlackwell2025

B200 192GB

See rental prices ↓

Current median: $6.25/hr across 4 providers (observed), 0.00 (0.0%) vs yesterday. Typical range: $6.08–$7.74/hr (P10–P90). 30-day trend unlocks after 30 days of captures

TDP 1000W · 3500 TFLOPS · 192GB VRAM · 8000 GB/s bandwidth

The B200 192GB has a 1000W TDP, 3500 TFLOPS of FP16 compute, 192GB of HBM3e VRAM, 8000 GB/s memory bandwidth.

Latest Blackwell data center GPU. FP4 support, massive VRAM, 2nd-gen Transformer Engine.

VRAM
192GB
HBM3e
FP16
3.5k
TFLOPS
Bandwidth
8.0k
GB/s
TDP
1000W
power
Best for:Next-gen LLM trainingUltra-large model inferenceScientific simulation

B200 192GB currently rents for a median of $6.25/hr across 4 providers (observed on-demand prices). B200 192GB has 192GB of memory. Compare its specs and listed cloud rental prices below. The workload must fit in VRAM with room for runtime memory; a listed price does not confirm stock.

B200 192GB Overview

The NVIDIA B200 192GB is a Blackwell data-center GPU built for frontier training and large-scale inference. Its 192GB of HBM3e, 3,500 TFLOPS of FP16/BF16 performance, and second-generation Transformer Engine represent a major step beyond Hopper, including lower-precision FP4 capability for compatible models.

With 8,000 GB/s of memory bandwidth, the B200 is designed to keep enormous model weights and KV caches moving at high throughput. The 192GB capacity can place very large quantized models or substantial context windows on one accelerator, while NVLink 5.0 at 1,800 GB/s supports tightly coupled multi-GPU scaling.

It is aimed at next-generation LLM training, ultra-large inference, and scientific workloads that can exploit a Blackwell cluster. The 1,000W power requirement, specialized fabric, and early supply constraints make it disproportionate for ordinary small-model inference.

Memory

VRAM192 GB
Memory TypeHBM3e
Bandwidth8000 GB/s
NVLink BW1800 GB/s

Compute Performance

FP645 TFLOPS
FP3280 TFLOPS
FP163500 TFLOPS
BF163500 TFLOPS
FP87000 TFLOPS
FP414000 TFLOPS
INT87000 TOPS

Hardware Specifications

Chip

ArchitectureGB202
GenerationBlackwell
Process NodeTSMC 4NP
Transistors208B
Die Size814 mm²
Release DateMarch 17, 2025

Processors

CUDA / Shader Cores18,432
Tensor Cores576

Clocks

Boost Clock1,800 MHz

Memory

VRAM192 GB
Memory TypeHBM3e
Memory Bus8192-bit
Bandwidth8000 GB/s
L2 Cache192 MB
NVLink BW1800 GB/s

Power

TDP1000 W
InterconnectNVLink 5.0 / PCIe 6.0

Relative Performance

FP16 Compute47%
VRAM Capacity67%
Mem Bandwidth50%

Relative to highest-spec GPU in database

Limitations

Very limited cloud availability — still ramping supply
Extremely high hourly cost vs H100/H200
Requires NVLink 5.0 infrastructure for multi-GPU setups

Live Cloud PricingOn-demand hourly rates

Live8 providers · -10% price spread
ProviderOn-DemandSpotRegionAvailability
BESTGPUniq
$9.42/hr—marketplaceunknown
$7.15/hr—Ashburn, VA, United States, Sydney, Australiaunknown
$6.25/hr—us-eastunknown
$6.99/hr—us-east-1unknown
$7.99/hr—USunknown
$3.50/hr—US/EU/APACunknown
$6.00/hr—US/EUunknown
$8.50/hr—eu-north1unknown
Cheapest / month
$6877
$9.42/hr × 730h
Median / month
$5833
at median price
Spot savings
N/A
vs on-demand (best)

Job Cost Estimator

At median $6.25/hr across 4 providers (observed on-demand). Spot pricing typically 40–70% lower.

7B LLM fine-tune
8× GPU, 24 h · Full fine-tune, single node
$1.2k
spot ~$540
70B LLM fine-tune
64× GPU, 72 h · Multi-node, 3-day run
$29k
spot ~$13k
Inference server
1× GPU, 730 h · 1 month continuous
$4.6k
spot ~$2.1k
Image gen batch
4× GPU, 8 h · Batch generation job
$200
spot ~$90

Estimates only. Actual cost depends on provider, region, availability, and utilization. Use the GPU cost calculator for custom inputs.

Compare B200 192GB vs…

Use Case Guidance

Next-gen LLM training
Ultra-large model inference
Scientific simulation

LLM Model Size Guidance

Max model (FP16)~96Bparameters at FP16 precision
Max model (INT8)~192Bparameters at INT8 precision
Max model (INT4)~384Bparameters at INT4/GGUF

Estimates only. Actual capacity depends on context length, KV cache, and framework overhead.

Related Guides

LLM APIs Running on This GPU Class

Providers that serve frontier LLM inference on Blackwell-class hardware.

Browse all 42 LLM models

B200 192GB vs Alternatives — Spec Comparison

SpecB200 192GB thisB200 SXM 192GBB200 NVL 192GBB300 SXM6
VRAM192GB HBM3e192GB HBM3e192GB HBM3e288GB HBM3e▲
Memory Bandwidth8000 GB/s▲8000 GB/s▲8000 GB/s▲8000 GB/s▲
FP16 TFLOPS3500▲3500▲3500▲2800
BF16 TFLOPS3500▲3500▲3500▲2800
FP8 TFLOPS7000▲7000▲7000▲5600
INT8 TOPS7000▲7000▲7000▲5600
TDP1000W▲1000W▲900W1000W▲
Process NodeTSMC 4NP▲TSMC 4NP▲TSMC 4NP▲TSMC 4NP▲
ArchitectureGB202GB202GB202GB300▲
Release Year2025▲2025▲2025▲2025▲
Max model (FP16)~96B params~96B params~96B params~144B params▲
Max model (INT4)~384B params~384B params~384B params~576B params▲
▲ indicates best value in row · FP16/BF16 TFLOPS at full precision · Max model estimates at 2 bytes/param (FP16) and 0.5 bytes/param (INT4)Full side-by-side comparison

Compare B200 192GB

Related GPUs

B200 192GB Frequently Asked Questions

What is the B200 192GB TDP?

The B200 192GB has a TDP (Thermal Design Power) of 1000W.

How many TFLOPS does the B200 192GB have?

The B200 192GB delivers 3500 TFLOPS of FP16 compute and 3500 TFLOPS BF16, and 7000 TFLOPS FP8.

How much VRAM does the B200 192GB have?

The B200 192GB has 192GB of HBM3e VRAM.

What is the B200 192GB's memory bandwidth?

The B200 192GB has a memory bandwidth of 8000 GB/s.

B200 192GB User Reviews

B200 192GB User Reviews

Loading reviews...

Ready to rent?

Compare B200 192GB prices across 130+ providers

Live on-demand & spot rates · monthly cost estimates · availability status

Compare rental prices