(All-cores)
(Single-core)
Specifications
Server Metadata
Vendor ID | gcp |
Server ID | 1723001 |
Name | a4-highgpu-8g |
Description | Accelerator Optimized: 8 NVIDIA B200 GPU, 224 vCPUs, 3968GB RAM |
Family | a4 |
Hw Virt | - |
Status | active |
Observed At | 2026-10-23T18:56:06.219385 |
Availability
| REGION / ID | SPOT | ONDEMAND |
|---|---|---|
Delhi (IN) / asia-south2 | 13.9192 USD/h | - |
St. Ghislain (BE) / europe-west1 | 21.2704 USD/h | - |
Ashburn (US) / us-east4 | 32.6064 USD/h | - |
Los Angeles (US) / us-west2 | 33.8184 USD/h | - |
Council Bluffs (US) / us-central1 | 39.6336 USD/h | - |
Moncks Corner (US) / us-east1 | 39.6336 USD/h | - |
Salt Lake City (US) / us-west3 | 41.3232 USD/h | - |
Dallas (US) / us-south1 | 41.6152 USD/h | - |
Jurong West (SG) / asia-southeast1 | 43.1384 USD/h | - |
Eemshaven (NL) / europe-west4 | 43.3896 USD/h | - |
Hamina (FI) / europe-north1 | 43.4984 USD/h | - |
Tokyo (JP) / asia-northeast1 | 52.1856 USD/h | - |
Processor
vCPUs | 224 |
CPU Allocation | Dedicated |
CPU Architecture | x86_64 |
System Resources and Accelerators
| MEMORY | |
|---|---|
Memory Amount | 4 TB |
| ACCELERATORS | |
|---|---|
Accelerator Count | 8 |
Accelerator Memory Min | 180 GiB |
Accelerator Memory Total | 1 TB |
Accelerator Type | GPU |
Accelerator Manufacturer | NVIDIA |
Accelerator Family | Blackwell |
Accelerator Model | B200 |
Accelerators |
|
| STORAGE | |
|---|---|
Storage Size | 12885 GB |
Storage Type | nvme ssd |
Storages |
|
| NETWORK | |
|---|---|
Inbound Traffic | 0 GB/month |
Outbound Traffic | 0 GB/month |
IPv4 | 0 |
Server Description
An accelerator-optimized platform featuring eight Blackwell GPUs, massive system memory, and high-capacity local NVMe SSD storage for intensive parallel workloads.
Google Cloud Platform a4-highgpu-8g is an accelerator-optimized server from the a4 family, designed for intensive parallel computing. It features 224 dedicated x86_64 vCPUs and 3968.0 GB of system memory, yielding 17.71 GB of RAM per core. The server is equipped with eight NVIDIA B200 Blackwell GPUs, providing a total of 1440 GB of VRAM. Local storage is supported by 12885 GB of NVMe SSD capacity. This high-density hardware configuration does not include complimentary public IPv4 addresses. Qualitatively, the server represents a premium hardware class with a high concentration of compute, memory, and accelerator resources. It is engineered for workloads such as large-scale machine learning model training, advanced data science tasks, large language model inference, and accelerated database operations.
Economics
Alternatives
Servers of the Same Family
| INSTANCE | vCPUs | MEMORY | Accelerators |
|---|
Similar Servers
| INSTANCE | VENDOR | vCPUs | MEMORY | Accelerators |
|---|