A2 family · Accelerator Optimizedx86_64 · Intel/AMD

a2-megagpu-16g

The a2-megagpu-16g machine type has 96 vCPUs and 1360 GB of memory. Pricing for this instance starts at $55.74 per hour and $40689.84 monthly in the us-central1 region.

Updated June 15, 2026
vCPUs
96
Memory
1360 GB
Network
Up to 100 Gbps
Storage
Local SSD

Save up to $20349/mo on every a2-megagpu-16g

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

All pricing options

us-central1 · USD · monthly = 730 hrs
Prices exclude local taxes
TermHourlyMonthlySavings
On-Demand
$55.7395$40689.84 Baseline
1-Year Committed Use Discount (CUD)
$35.1155$25634.30 -37%
3-Year Committed Use Discount (CUD)
$19.5092$14241.72 -65%
Preemptible / Spot
$27.8638$20340.60 -50%
You could save $20349.23/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
$40689.84
730 hrs × $55.7395/hr × 1

Compute

x86_64 · Intel/AMD
vCPUs i96
Memory i1360 GB
Physical processorIntel Cascade Lake
Nested VirtualizationNot supported
Sole TenantNot supported
GPU16 GPU
Within A2 family
a2-highgpu-1g
12 vCPU
a2-ultragpu-1g
12 vCPU
a2-highgpu-2g
24 vCPU
a2-ultragpu-2g
24 vCPU
a2-highgpu-4g
48 vCPU
a2-ultragpu-4g
48 vCPU
a2-highgpu-8g
96 vCPU
a2-ultragpu-8g
96 vCPU
a2-megagpu-16g
96 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 bandwidth100 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.
Speca2-megagpu-16ga2-highgpu-1gΔ
vCPUs9612-87%
Memory1360 GB85 GB-94%
Hourly Price$55.7395$3.6734-93%
Monthly Price$40689.84$2681.57-93%
Cost optimization

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

Connect your GCP account. We'll identify every underutilized a2.* instance and show exactly where downsizing, cleaning up orphaned disks, or shifting to CUDs saves money — with zero code changes.