Free cloud & LLM pricing tools — AWS EC2 + LLM API Pricing Explorer

GCP Compute Engine pricing glossary

The pricing model behind every number in the GCP explorer, in plain English. GCP prices compute differently from AWS and Azure — per vCPU and per GB, with automatic sustained-use discounts — so start here if the terms are new.

On-Demand (per-second billing)

Pay-as-you-go Compute Engine pricing, billed per second after a one-minute minimum — the baseline every discount is measured against.

Per-vCPU / per-GB billing

GCP prices compute and memory as separate resources, so a machine price is vCPUs × core rate + GB × memory rate.

Sustained Use Discount (SUD)

An automatic discount for running eligible VMs for a large share of the month — up to 20% on N2, N2D and C2, applied with no commitment.

Committed Use Discount (CUD)

A 1- or 3-year commitment for a large discount — resource-based CUDs are tied to a region and machine family; flexible CUDs trade some discount for portability.

Spot VMs (Preemptible)

Spare capacity at a deep, variable discount — Compute Engine can preempt (reclaim) the VM at any time with 30 seconds’ notice.

Custom machine types

Pick your own vCPU and memory instead of a predefined bundle — priced by the same per-vCPU and per-GB rates.

Extended memory

Memory added to a custom machine beyond the standard per-vCPU ceiling, billed at a higher rate.

Machine series (E2, N-series, C-series, M-series)

GCP groups VMs into families with different price/performance: E2 cost-optimized, N general-purpose, C compute-optimized, M memory-optimized.

Axion & Arm (C4A)

C4A runs on Google’s Axion Arm processor — strong price/performance for scale-out and cloud-native workloads that run on Arm.

Sole-tenant nodes

Dedicated physical servers that host only your VMs — for licensing or compliance — at a premium over shared tenancy.

Effective price at 100% monthly use

The on-demand rate after the maximum sustained-use discount — what a VM that runs the whole month effectively pays.

Normalized pricing ($/vCPU, $/GB)

On-demand price divided by vCPU or by GB of memory — on GCP these are the literal billing units.

Max saving %

The deepest discount versus on-demand across sustained-use and committed-use discounts for a machine.

Best price region

The cheapest launch-set region for a machine and the saving versus the region you are viewing.

Other glossaries: AWS EC2 pricing →Azure VM pricing →