Interactive 3D Model
Loading viewer...
Drag to rotate ┬╖ Scroll to zoom
ARCHITECTURE
Hopper
SERIES
H200
Memory
141GB HBM3e
Bandwidth
4.8 TB/s
TDP
700W
In StockH200┬╖Best for: Large Model Training
NVIDIA H200 Tensor Core GPU
Hopper
|Data Center GPUsOverview
The NVIDIA H200 builds on the Hopper architecture with 141GB of HBM3e memory and 4.8 TB/s bandwidth. Designed for large-scale AI models and inference workloads, the H200 delivers up to 2x performance gain over the H100 in memory-bound scenarios.
Key Features
- 141GB HBM3e
- 4.8 TB/s Bandwidth
- Transformer Engine
Ideal Use Cases
ai trainingai inference
Full Specifications
Complete technical specifications for the H200.
Memory
- Memory
- 141GB HBM3e
- Memory Bandwidth
- 4.8 TB/s
Compute
- Tensor Cores
- 528
- CUDA Cores
- 18432
- FP8 TFLOPS
- 1979
- FP16 TFLOPS
- 989
- TF32 TFLOPS
- 494.5
- FP64 TFLOPS
- 60
Interconnect
- Interconnect
- NVLink 4.0, 900GB/s
Physical
- TDP
- 700W
- Form Factor
- SXM5
- Cooling
- Liquid Cooled
Manufacturing
- Launch Date
- Q2 2024
- Manufacturing Process
- 4nm