Compute Comparison
NVIDIAHopper2024

H200 141GB

H100 die with HBM3e memory upgrade. Same compute, 76% more VRAM, 43% more bandwidth than H100.

VRAM
141GB
HBM3e
FP16
2.0k
TFLOPS
Bandwidth
4.8k
GB/s
TDP
700W
power
Best for:70B+ model inferenceMemory-bound LLM servingLarge context windows

H200 141GB Overview

The NVIDIA H200 141GB is an upgraded variant of the H100 that replaces the 80GB HBM3 memory with 141GB of HBM3e, increasing both capacity and bandwidth. It uses the same GH100 die and Hopper architecture as the H100, delivering identical compute throughput: 1,979 TFLOPS FP16/BF16 and 3,958 TFLOPS FP8. The key difference is memory: 141GB at 4,800 GB/s bandwidth — 76% more capacity and 43% more bandwidth than the H100 80GB.

The 141GB HBM3e configuration fits a 70B parameter model at FP16 on a single card (140GB required) — something the H100 80GB cannot do without quantization. For 70B inference, this eliminates the need for tensor parallelism across two H100s, reducing latency and infrastructure complexity. The 4,800 GB/s bandwidth directly improves tokens-per-second for memory-bound inference: a 70B model at FP16 generates approximately 43% more tokens per second on H200 vs H100 at the same batch size.

The H200 is the right choice when running 70B+ models at FP16 on a single card, when memory bandwidth is the primary bottleneck for inference throughput, or when large context windows (>128K tokens) require more than 80GB of KV cache. It is not the right choice when compute is the bottleneck (same TFLOPS as H100), when models fit comfortably in 80GB, or when cost is the primary concern — H200 commands a 40–60% premium over H100 with no compute improvement.

Memory

VRAM141 GB
Memory TypeHBM3e
Bandwidth4800 GB/s
NVLink BW900 GB/s

Compute Performance

FP3267 TFLOPS
FP161979 TFLOPS
BF161979 TFLOPS
FP83958 TFLOPS
INT83958 TOPS

Hardware

ArchitectureGH100
GenerationHopper
Process NodeTSMC 4N
Transistors80B
TDP700 W
InterconnectNVLink 4.0 / PCIe 5.0
Release Year2024

Relative Performance

FP16 Compute26%
VRAM Capacity49%
Mem Bandwidth30%

Relative to highest-spec GPU in database

Limitations

Premium pricing over H100 — ~40–60% higher hourly cost
Limited availability vs H100 — fewer providers offer it
Same compute as H100 — only memory bandwidth improvement

Live Cloud PricingOn-demand hourly rates

Loading live prices…

Compare H200 141GB vs…

Use Case Guidance

70B+ model inference
Memory-bound LLM serving
Large context windows

LLM Model Size Guidance

Max model (FP16)~70Bparameters at FP16 precision
Max model (INT8)~141Bparameters at INT8 precision
Max model (INT4)~282Bparameters 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 Hopper-class hardware.

Browse all 42 LLM models Self-hosting guide: best LLM models for this GPU

Related GPUs

Frequently Asked Questions

How much VRAM does the H200 141GB have?

The H200 141GB has 141GB of HBM3e memory with 4800 GB/s bandwidth. This enables running models up to approximately 282B parameters at INT4 precision, 141B at INT8, or 70B at FP16.

What is the FP16 performance of the H200 141GB?

The H200 141GB delivers 1979 TFLOPS of FP16 performance and 1979 TFLOPS BF16, and 3958 TFLOPS FP8. INT8 throughput is 3958 TOPS. For transformer inference, memory bandwidth (4800 GB/s) is often the binding constraint rather than raw TFLOPS.

What is the H200 141GB best used for?

The H200 141GB is best suited for: 70B+ model inference, Memory-bound LLM serving, Large context windows. H100 die with HBM3e memory upgrade. Same compute, 76% more VRAM, 43% more bandwidth than H100.

What interconnect does the H200 141GB use?

The H200 141GB uses NVLink 4.0 / PCIe 5.0 with 900 GB/s NVLink bandwidth for multi-GPU configurations. NVLink enables near-linear tensor-parallel scaling across multiple cards for models that exceed single-card VRAM.

What LLM model sizes can the H200 141GB run?

With 141GB of HBM3e, the H200 141GB can run models up to approximately 70B parameters at FP16 (2 bytes/param), 141B at INT8 (1 byte/param), or 282B at INT4/GGUF (0.5 bytes/param). These are estimates — actual capacity depends on context length, KV cache size, and framework overhead. Longer context windows require more KV cache memory, reducing the effective model size that fits.

How does the H200 141GB compare to the A100 for LLM inference?

The H200 141GB has 1979 TFLOPS FP16 vs the A100 80GB's 312 TFLOPS, and 4800 GB/s memory bandwidth vs the A100's 2,039 GB/s. For memory-bound autoregressive LLM inference, bandwidth is the primary determinant of tokens-per-second. The H200 141GB's higher bandwidth gives it a throughput advantage for large model inference.

What is the power consumption of the H200 141GB?

The H200 141GB has a TDP (Thermal Design Power) of 700W. This is the maximum sustained power draw under full load. For data center deployments, total rack power consumption is typically 1.2–1.5× the GPU TDP when accounting for CPU, memory, networking, and cooling overhead. At 700W, the H200 141GB is in the high-power tier — requires specialized data center infrastructure with high-density power delivery.

Ready to rent?

Compare H200 141GB prices across 97+ providers

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

Compare rental prices