How much does an idle g5.12xlarge cost on AWS?

A g5.12xlarge billed on-demand and left running around the clock costs about $4,141/mo at its cheapest listed region (us-east-1). Scheduling it off outside business hours cuts that to roughly $1,232/mo — a $2,908/mo saving.

Cheapest on-demand

$5.6720/hr

us-east-1

Running 24/7

$4,141/mo

730-hour month

Business-hours savings

$2,908/mo

70% off the 24/7 bill

vCPU 48 Memory 192 GB GPU 4× NVIDIA A10G Family g5 Provider AWS

Price by region

Region Hourly (on-demand) Spot (typical) ≈ Monthly (24/7)
us-east-1 Cheapest $5.6720/hr $2.7793/hr ≈ $4,141/mo
us-east-2 $5.6720/hr $2.2121/hr ≈ $4,141/mo
us-west-2 $5.6720/hr $2.6091/hr ≈ $4,141/mo
eu-west-1 $6.3317/hr $3.0392/hr ≈ $4,622/mo
ap-south-1 $6.8109/hr $4.0184/hr ≈ $4,972/mo
eu-central-1 $7.0928/hr $3.4046/hr ≈ $5,178/mo
ap-southeast-2 $7.3747/hr $1.9174/hr ≈ $5,384/mo
ap-northeast-1 $8.2261/hr $3.7017/hr ≈ $6,005/mo
sa-east-1 $9.6413/hr $5.3991/hr ≈ $7,038/mo

What a schedule saves

Schedule Monthly cost Savings vs 24/7
Running 24/7 $4,141/mo baseline
Weekdays only (off on weekends) $2,958/mo -$1,183/mo (29%)
Business hours only (10×5) $1,232/mo -$2,908/mo (70%)

Assumes a 730-hour month at on-demand list price. Actual savings depend on when the instance is actually used.

About the g5 family

G5 steps up to NVIDIA A10G GPUs, aimed at heavier ML inference, mid-size training, and graphics workloads that outgrow the T4. Teams reach for it when a g4dn box stops keeping up, but the usage pattern doesn't change: active during working hours, fully idle the rest of the week — and at a higher hourly rate, every idle hour costs more.

Regional pricing varies: g5.12xlarge runs $5.6720/hr in us-east-1, the cheapest region priced here, versus $9.6413/hr in sa-east-1, the priciest — a 41% spread across regions for the identical instance type.

The next size up in the same family, g5.16xlarge (64 vCPU / 256 GB RAM), costs about $2,990/mo running 24/7 at its cheapest region — roughly 28% less than g5.12xlarge's ~$4,141/mo — for 1.3x the vCPU. Note the trade: it carries 1 GPU versus 4 on the g5.12xlarge — cheaper only if your work isn't GPU-bound.

Common questions

How much does an idle g5.12xlarge cost per month?

At the cheapest listed region (us-east-1), a g5.12xlarge costs $5.6720/hr, or about $4,141/mo if it runs 24/7. That's the on-demand list price — an idle instance costs exactly the same as a busy one, since cloud billing doesn't know the difference.

How much can I save by scheduling it off outside work hours?

Restricting it to business hours only (10×5) saves about $2,908/mo (70%) versus running 24/7. A lighter schedule — on around the clock on weekdays, off on weekends — still saves about $1,183/mo (29%).

When is g5 worth the premium over g4dn?

When the model or batch size outgrows the T4's 16 GB of GPU memory or its throughput — the A10G raises GPU memory to 24 GB and roughly doubles the compute. If a g4dn box is merely slow at training runs you do twice a week, scheduling beats upgrading: premium hardware only pays for itself while it is actually busy.

Other g5 sizes

Stop paying for idle AWS servers

Idlefy runs a free audit of your fleet, then flips dev servers to stopped-by-default: lease one when you need it, and when the lease ends it shuts itself off — stop, not terminate, persistent disks intact. Timers are set by your team, never guessed, and Idlefy only manages instances tagged idlefy=enabled — on AWS that boundary is enforced by the IAM policy itself.

Run a free idle audit

Prices are on-demand list price for Linux instances and exclude storage, network, data transfer, and any negotiated or volume discounts. Actual bills vary by usage and account-level pricing. Prices as of 2026-07-20. Spot figures are indicative — AWS values are derived from AWS's published typical savings vs on-demand for the region; GCP values are the listed Spot price. Spot capacity can be reclaimed by the provider. Stopping preserves EBS and persistent-disk volumes; data on local instance-store or local-SSD drives does not survive a stop.