GCP Compute Instance

a3-megagpu-8g

The a3-megagpu-8g is a GPU-accelerated GCP Compute instance with 208 vCPUs and 1872 GB of memory, 8 H100-MEGA GPUs, powered by Intel. Ideal for deep learning training, high-performance computing, and GPU-intensive inference workloads. On-demand pricing starts at $9.4601/hr in us-central1.

Price History

Pricing by Region

Region On-Demand Spot Reserved 1Y
us-central1$9.460$3.863$6.884
us-west1$9.460$3.863$6.884
europe-north1$10.416$1.538$7.244
asia-east1$10.954$3.667$7.618
us-east4$92.214$18.527$63.622
us-east5$93.409$26.023$64.087
europe-west1$101.537$29.629$69.906
us-west4$103.861$20.727$71.657
europe-west3$108.813$21.717$75.117
australia-southeast1$115.268$30.014$79.740
europe-west4$117.514$34.304$82.930
asia-northeast1$118.422$23.098$81.750
asia-southeast1$119.251$38.121$82.098

Frequently Asked Questions

What is the a3-megagpu-8g?
The a3-megagpu-8g is a GPU-accelerated GCP Compute instance with 208 vCPUs and 1872 GB of memory, 8 H100-MEGA GPUs, powered by Intel. Ideal for deep learning training, high-performance computing, and GPU-intensive inference workloads. On-demand pricing starts at $9.4601/hr in us-central1.
What is the cheapest region for a3-megagpu-8g?
us-central1 at $9.460/hr on-demand.
How much does a3-megagpu-8g cost per month?
Starting at $6905.84/mo on-demand in us-central1, or $1122.40/mo with spot pricing.

Run a free assessment to identify overprovisioned workloads, idle capacity, and your potential savings, in minutes.

Most clusters are overprovisioned.
Let's prove yours is.