Cloud VM Price Comparison: AWS vs Azure vs GCP

Find the cheapest virtual machine for your workload across AWS, Microsoft Azure and Google Cloud, side by side, in seconds. Tell it how many vCPUs and how much memory you need, pick a location, and get the cheapest matching instance on each cloud, ranked, with hourly and monthly prices.

Prices come from each provider's official public price lists (AWS Price List, Azure Retail Prices API, Google Cloud Billing Catalog) and are refreshed daily.

What you get

For 4 vCPU / 16 GB, Linux, Canada (real output, October 2026):

Cloud	Instance	Arch	$/hour	$/month	vs cheapest
GCP	e2-standard-4	x64	0.1475	107.71	cheapest
Azure	Standard_D4ps_v6	Arm	0.1560	113.88	+5.7%
AWS	m6g.xlarge	Arm	0.1712	124.98	+16.0%

Intel pricing is nearly identical across the three clouds. The real savings usually come from AMD and Arm instances, and this Actor finds them for you.

Why use it

• Three clouds, one answer. No more switching between three pricing calculators that each name instances differently.
• Matches by spec, not by name. You ask for vCPU and memory; it finds the cheapest instance on each cloud that meets them, and flags when the best option is larger than you asked for.
• Includes Google Cloud. GCP prices most machines per vCPU and per GB of RAM rather than per instance; this Actor does that math for you.
• Batch mode. Compare a whole server inventory in one run, for migration planning, FinOps reviews or client proposals.
• Fair comparisons. Burstable instances (AWS t-series, Azure B-series, GCP shared-core e2) are excluded by default, because their throttled CPUs aren't comparable to dedicated ones. Turn them on if you want them.

Who it's for

• FinOps and cloud cost teams checking whether a workload would be cheaper elsewhere
• Consultants and MSPs preparing multi-cloud proposals and migration estimates
• Engineers choosing where to run a new service
• AI agents answering "which cloud is cheapest for X?" (callable through the Apify MCP server)

How to use it

1. Enter the vCPUs and memory (GB) you need.
2. Choose Linux or Windows (Windows prices include the Windows Server license).
3. Pick a location. Presets compare the nearest region on each cloud:

Preset	AWS	Azure	GCP
Canada Central	ca-central-1	canadacentral	northamerica-northeast2
Canada East	ca-central-1	canadaeast	northamerica-northeast1
US East	us-east-1	eastus	us-east4
US Central	us-east-2	centralus	us-central1
US West	us-west-2	westus2	us-west1

Or choose Custom regions and enter any region code per cloud, for example {"aws": "eu-west-1", "azure": "northeurope", "gcp": "europe-west1"}. Leave a cloud out to skip it.

4. Optionally restrict CPU architecture to x64 if your software can't run on Arm, or raise options per cloud to see the top few choices from each provider.

Batch mode

To compare many workloads in one run, fill in Batch: multiple workloads:

[
  {"name": "web", "vcpu": 2, "memoryGb": 8},
  {"name": "api", "vcpu": 4, "memoryGb": 16},
  {"name": "db", "vcpu": 8, "memoryGb": 64, "os": "Windows"}
]

Output

One row per recommended instance, in a table you can export as CSV, Excel or JSON.

Field	Meaning
workload	Workload name (from batch mode, or workload-1)
provider	aws, azure or gcp
instanceType	Instance or VM size name
vcpu, memoryGb	Instance specs
arch	x64 or arm64
os	Linux or Windows
region	Region code
priceHourUsd, priceMonthUsd	On-demand price (month = 730 hours)
vsCheapestPct	How much more it costs than the cheapest option, in %
isCheapest	true for the cheapest option for this workload
exactMatch	false when the instance is larger than requested
pricesUpdatedAt	When that cloud's prices were last refreshed

Pricing

$0.02 per comparison. One comparison is one workload compared across the clouds, however many options per cloud you ask for. A 100-server inventory costs $2. Workloads with no match on any cloud are not charged.

Coverage

• Pricing model: on-demand (pay-as-you-go), USD, compute only. Disks, network and licenses other than Windows Server are not included.
• AWS: all EC2 instance families in every commercial and GovCloud region, plus Local Zones.
• Azure: D-series v5/v6 (general purpose, including AMD and Arm), E-series v5/v6 (memory optimized), Fs v2 (compute optimized) and B-series (burstable), in all public regions.
• Google Cloud: E2, N2, N2D, C3 and C4 families in all regions.

Specialized families (GPU, HPC, very large memory) and reserved, savings-plan and spot pricing are on the roadmap.

FAQ

How current are the prices? They are refreshed from the providers' official price lists every day. Each result includes pricesUpdatedAt.

Why didn't I get an exact match? Not every cloud offers every vCPU/memory combination. The Actor picks the cheapest instance that meets or exceeds your request and sets exactMatch to false, so you can see when you'd be paying for extra capacity.

Why is Arm often the cheapest? AWS Graviton, Azure Cobalt and Ampere instances are typically priced 10–20% below comparable Intel instances. If your software (or its container images) only runs on x86, set CPU architecture to x64.

Are the vCPUs comparable across clouds? Close, but not identical. A vCPU is a hardware thread on most Intel and AMD instances, while Arm instances usually give a full physical core per vCPU. For performance-sensitive workloads, benchmark before committing.

Do the prices include discounts? No. These are public list prices. Committed-use discounts, savings plans, enterprise agreements and GCP sustained-use discounts can lower your actual bill.

Using it from code or AI agents

Call it through the Apify API or any Apify client, or connect it to Claude, ChatGPT, Cursor and other AI tools through the Apify MCP server. Ask "what's the cheapest 8 vCPU / 32 GB Linux VM in US East?" and let the agent run it.

Feedback

Missing a region, instance family or feature you need? Open an issue on this Actor's Issues tab.