H200
141GB HBM3e per GPU
View GPU All GPUsH100
Hopper architecture80GB SXM5 or 94GB NVL per GPU
GPU Catalog / NVIDIA
H100
GPU infrastructure.
A Hopper data-center GPU family for transformer training, large-scale inference, and scientific computing.
- Architecture
- Hopper
- GPU memory
- 80GB SXM5 or 94GB NVL per GPU
- Published from
- $3.00 / hr
Published pricing
Available configurations.
Current Geodd hourly rates. GPU memory totals reflect the accelerator count in each published configuration.
| Configuration | Variant | Total GPU memory | Hourly rate | Action |
|---|---|---|---|---|
1x GPUgpu_1x_h100 | H100 SXM5 | 80GB HBM380GB HBM3 each | $3.00per hour | Deploy |
2x GPUgpu_2x_h100 | H100 SXM5 | 160GB HBM380GB HBM3 each | $6.00per hour | Deploy |
4x GPUgpu_4x_h100 | H100 SXM5 | 320GB HBM380GB HBM3 each | $12.00per hour | Deploy |
8x GPUgpu_8x_h100 | H100 SXM5 | 640GB HBM380GB HBM3 each | $24.00per hour | Deploy |
Hardware profile
Where H100 fits.
H100 SXM5 and H100 NVL are distinct variants. SXM5 uses 80GB HBM3 per GPU, while NVL uses 94GB HBM3 per GPU, so Geodd identifies the variant and memory total separately for every published configuration.
Choose H100 when training or inference throughput is the priority and the workload fits the memory envelope of the selected SXM5 or NVL variant.
NVIDIA product information- 01
Transformer training
Train and fine-tune transformer models on Hopper-generation hardware designed for modern AI training workloads.
- 02
Deep-learning inference
Serve large neural networks and language models on a data-center GPU designed for sustained inference compute.
- 03
Scientific and HPC workloads
Use H100 for GPU-accelerated scientific applications that require double-precision computation alongside AI capability.
Deployment sizing
Size before you provision.
Hardware fit depends on the model, precision, batch size, software stack, and total working set. Published use cases are not performance guarantees.
- 01 / Memory
Confirm the working set
Include weights, caches, activations, batches, and framework overhead.
- 02 / Software
Validate the stack
Check CUDA, driver, framework, precision, and container requirements.
- 03 / Capacity
Confirm provisioning
Published pricing does not guarantee immediate capacity in every configuration.