GCP Pricing Calculator

Google Cloud Compute Engine machine types, pricing models, discounts, and cost-optimization strategies — all in one place.

GCP Compute Engine Machine Families

Google Cloud organizes Compute Engine into distinct machine families, each optimized for different workload profiles. Choosing the right family is the single biggest lever for cost optimization — differences of 30–50% in hourly rate are common between families for equivalent resource counts.

E2 (Economy): Best price-performance for general workloads. Uses a shared-core model with dynamic resource allocation. No Sustained Use Discounts, but very low baseline cost. Great for web servers, small databases, and development environments.

N2 / N2D (General Purpose): Balanced performance for a broad range of workloads. N2 uses Intel Cascade Lake processors, N2D uses AMD EPYC. Both support Sustained Use Discounts and Committed Use Discounts. Recommended for production workloads needing consistent performance.

C2 / C2D (Compute-Optimized): Highest per-core performance on GCP. Ideal for gaming, HPC, ad serving, and CPU-intensive analytics. Higher hourly cost but fewer cores needed for the same throughput. C2D (AMD) offers slightly better price-performance than C2 (Intel) for vectorized workloads.

T2D (Scale-Out): AMD EPYC Milan processors with excellent single-threaded performance and competitive pricing. Best for scale-out workloads like web serving, containerized microservices, and media transcoding.

Common Machine Type Pricing (us-central1, on-demand)

Prices are approximate and subject to change. Always verify current pricing at cloud.google.com/compute/all-pricing. Spot VM prices fluctuate based on regional capacity.

Pricing Models: On-Demand, Spot, and Committed Use

GCP offers multiple pricing models that can dramatically reduce your compute bill. Understanding when to use each is the key to cloud cost optimization.

• On-Demand: Full price, no commitment. Best for unpredictable workloads, new projects, or short-term needs. No upfront payment.
• Spot VMs: Up to 91% discount vs on-demand. VMs can be preempted with 30 seconds notice. Ideal for batch jobs, CI/CD pipelines, data processing, and fault-tolerant workloads.
• Committed Use Discounts (1-year): ~37% off on-demand. No upfront payment required. Best for stable production workloads running 24/7.
• Committed Use Discounts (3-year): Up to 55% off on-demand. Requires a 3-year commitment on resource spend in a region. Maximum savings for long-term infrastructure.
• Sustained Use Discounts (automatic): Applied automatically for N1/N2/N2D VMs running 25%+ of the month. Scales up to 30% for full-month usage. No action required.

GKE Pricing Breakdown

Google Kubernetes Engine has a straightforward pricing model: you pay for the cluster management fee plus the underlying Compute Engine nodes.

Cloud Run Pricing

Cloud Run is GCP's fully managed serverless container platform. You pay only for what you use — there is no charge when your service is not handling requests.

• CPU: $0.00002400/vCPU-second (billed only during request processing)
• Memory: $0.00000250/GB-second
• Requests: $0.40 per million requests
• Free tier: 2 million requests/month, 360,000 vCPU-seconds, 180,000 GB-seconds — enough for many small services at zero cost
• CPU always-on option: CPU is allocated for the full instance lifetime (even between requests), allowing background work. Priced at a slight premium but with always-allocated CPU charges.

For a service handling 10 million requests/month, each taking 200ms with 1 vCPU and 256 MB RAM, the estimated Cloud Run cost is approximately $80–100/month — often significantly cheaper than a dedicated VM that sits idle between traffic spikes.

Cost Optimization Strategies

• Right-size before committing: Use Cloud Monitoring to identify underutilized VMs and right-size them before purchasing CUDs.
• Mix Spot and on-demand: Use managed instance groups with a mix of Spot VMs (80%) and on-demand (20%) for fault-tolerant workloads to save 70%+.
• Use regional commitments: CUDs apply to any matching VM in a region — you can upgrade, downgrade, or change machine types without losing the discount.
• Enable SUD automatically: N2 and N2D machines accrue Sustained Use Discounts automatically. No configuration needed — just avoid restarting VMs mid-month unnecessarily.
• Consider Cloud Run for APIs: Variable-traffic APIs and microservices are almost always cheaper on Cloud Run than on always-on VMs at moderate request volumes.
• Use committed use for GKE nodes: GKE node pools run on Compute Engine — all VM discounts apply. Apply CUDs to the node machine types for maximum savings.