Grace Hopper superchip combining ARM CPU with H100 GPU.
Weekly median price per GPU per hour · Get the data
By provider shows one card per company, placed where its best offer ranked. By configuration lists every offer.
What you can rent comes first: in stock, then waitlist, then not reported, then out of stock. Priced offers rank ahead of quote-only, and on-demand ahead of other billing types.
Within a group, five factors set the order:
We match the provider, its country, GPU model, form factor, billing type, availability and instance name. Matching is partial and case-insensitive. Everyday words work too, so "interruptible" finds spot and "sold out" finds out of stock.
No providers match .
Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Verify before provisioning. More on how we price.
One GH200 has 96 GB of VRAM. In practice, that's enough memory for roughly 146B parameters at 4-bit or 38B at 16-bit, assuming a 32K context. Below are some open-weight LLMs, with the estimated memory and GPUs each one needs.
| Model | Memory (INT4 / FP4) | GH200s needed | Cost /hr | Cost /mo |
|---|---|---|---|---|
Qwen3.8-27B
27B
|
18 GB
|
1
|
$3.87
|
$2,786
|
Gemma 4 31B
30.7B
|
22 GB
|
1
|
$3.87
|
$2,786
|
GPT-OSS-120B
117B · 5.1B active
|
67 GB
|
1
|
$3.87
|
$2,786
|
GLM-5.3-Flash
320B · 18B active
|
178 GB
|
3
|
–
|
–
|
MiniMax-M3
428B · 23B active
|
239 GB
|
3
|
–
|
–
|
DeepSeek V4.1 Flash
552B · 16B active
|
306 GB
|
4
|
–
|
–
|
Kimi K3
2.8T · 104B active
|
1,544 GB
|
18
|
–
|
–
|
Estimates based on the median on-demand rate. Memory is weights plus FP8 KV cache at 32K context per request (FP16/BF16 in the 16-bit column). GPU counts assume 90% of advertised VRAM is usable. No guarantee of runtime support, usable performance, or that a matching quantized build exists. Pricing methodology.
Nvidia GH200 · Per GPU
| Compute · dense | |
|---|---|
| FP8 | 1,979 TFLOPS 3,958 with sparsity |
| FP16 / BF16 | 989.5 TFLOPS 1,979 with sparsity |
| INT8 | 1,979 TOPS 3,958 with sparsity |
| FP32 | 67 TFLOPS |
| FP64 | 34 TFLOPS |
| Precision support | FP8FP16BF16TF32FP32FP64INT8 |
| Memory | |
|---|---|
| Capacity | 96 GB HBM3 |
| Bandwidth | 4,000 GB/s |
| ECC | Yes |
| Silicon | |
|---|---|
| Architecture | Hopper |
| Process | TSMC 4N |
| Transistors | 80 billion |
| Shader cores | 16,896 CUDA cores |
| Matrix cores | 528 Tensor cores |
| Compute units | 132 SMs |
| Fabric and host | |
|---|---|
| GPU interconnect | NVLink 900 GB/s |
Source: official Nvidia GH200 datasheet.
As of October 23, 2026, the median on-demand price is $3.87 per GPU per hour across 3 providers with a priced on-demand config.
| Billing type | Configs | Median /GPU/hr | Cheapest |
|---|---|---|---|
On-demand | 5 | $3.87 | $2.52 (in stock, Sesterce) |
Spot | 1 | Sold out |
The cheapest verified in-stock on-demand GH200 tracked by GetDeploying is $2.52 per GPU per hour from Sesterce.
At 720 hours per month, one GH200 costs an estimated $2,786 at the median on-demand price. Cheapest verified in stock: $1,814 per month on-demand.
GetDeploying currently tracks GH200 configs from 5 providers. Sesterce has verified in-stock on-demand configs. See the full price comparison above for every provider and config.
One GH200 runs models up to roughly 146B parameters at 4-bit quantization or 38B at 16-bit, assuming a 32K context. Larger models run across multiple GPUs: the model table above shows the estimated memory and GPU count for popular open-weight LLMs.
Grace Hopper superchip combines ARM CPU with H100 GPU and 96GB or 144GB HBM3e. Unified memory architecture with 900 GB/s NVLink. Efficient for AI workloads that benefit from tight CPU-GPU integration.
ARM CPU ecosystem may have compatibility issues with some software stacks. For standard GPU cloud workloads, a regular H100 is simpler to deploy and more widely available.
Last updated
