A3 family · Accelerator Optimizedx86_64 · Intel/AMD

a3-megagpu-8g

The a3-megagpu-8g machine type has 208 vCPUs and 1872 GB of memory. Pricing for this instance starts at $119.28 per hour and $87071.75 monthly in the asia-northeast1 region.

Updated June 16, 2026
vCPUs
208
Memory
1872 GB
Network
Up to 1600 Gbps
Storage
Local SSD

Save up to $70111/mo on every a3-megagpu-8g

Connect Google Cloud — we'll find the optimal pricing model for your workload.

All pricing options

asia-northeast1 · USD · monthly = 730 hrs
Prices exclude local taxes
TermHourlyMonthlySavings
On-Demand
$119.2764$87071.75 Baseline
1-Year Committed Use Discount (CUD)
$82.0694$59910.69 -31%
3-Year Committed Use Discount (CUD)
$51.9596$37930.54 -56%
Preemptible / Spot
$23.2336$16960.49 -81%
You could save $70111.26/mo on this instance with the right pricing model.
Find my savings

Monthly cost estimator

Estimate your spend based on actual usage.
730 h
1h365h730h (24/7)
1
Estimated monthly
$87071.75
730 hrs × $119.2764/hr × 1

Compute

x86_64 · Intel/AMD
vCPUs i208
Memory i1872 GB
Physical processorIntel Sapphire Rapids
Nested VirtualizationNot supported
Sole TenantNot supported
GPU8 GPU
Within A3 family
a3-highgpu-1g
26 vCPU
a3-highgpu-2g
52 vCPU
a3-highgpu-4g
104 vCPU
a3-megagpu-8g
208 vCPU
a3-highgpu-8g
208 vCPU

Lock in this rate across your fleet — typically save 30–40%

Connect your Google Cloud account to apply this pricing logic to every running instance automatically.

Networking

Max egress bandwidth1600 Gbps
Tier 1 bandwidthN/A
Enhanced networking iYes (gvnic/virtio)
IPv6 supportYes (Dual-stack VPC)

Storage

Max persistent disks128
Max disk size512 TB
Local SSDsSupported (Scratch Disk)

Compare with another instance

vs.
Speca3-megagpu-8ga3-highgpu-1gΔ
vCPUs20826-87%
Memory1872 GB234 GB-87%
Hourly Price$119.2764$15.8304-87%
Monthly Price$87071.75$11556.21-87%
Cost optimization

GCP recommender alerts can cut this bill ~30-40%.

Connect your GCP account. We'll identify every underutilized a3.* instance and show exactly where downsizing, cleaning up orphaned disks, or shifting to CUDs saves money — with zero code changes.