Cloud Compute Pricing Comparator
Enter a target vCPU and RAM size, see the closest matching general-purpose instance on AWS, Azure and GCP with an approximate hourly and monthly cost side by side. Useful for a rough gut-check before you commit to a provider, not a substitute for each provider's actual pricing calculator when you're about to sign off on real spend.
Why the numbers here are "indicative," and why that matters
Cloud pricing changes, varies by region, and on-demand list price is genuinely the most expensive way to run anything long-term — this tool uses on-demand Linux rates because they're the only number that's directly comparable across providers without picking a specific commitment term. If you're actually budgeting a production workload, run the same numbers through the provider's own calculator with your actual region and commitment plan before treating anything here as a real quote.
The pricing lever that matters more than which provider you pick
On-demand is list price — full flexibility, no commitment, and the most expensive per hour by a wide margin. Reserved instances or savings plans (1 or 3 year commitments) typically cut 30–60% off on-demand. Spot instances (AWS/Azure) or preemptible VMs (GCP) can cut 60–90%, in exchange for the provider being able to reclaim the instance with short notice — genuinely fine for batch jobs, CI runners and fault-tolerant workloads, genuinely wrong for anything that can't tolerate an unscheduled interruption. Comparing on-demand prices across providers is a reasonable first pass, but the actual savings available within a single provider from choosing the right commitment model are usually bigger than the difference between providers at list price.
Specs match, real-world performance doesn't always
A "2 vCPU / 8GB" instance on three different providers is not guaranteed to perform identically — vCPU definitions differ (some are full physical cores, some are hyperthreads), and network and storage throughput are frequently separate line items entirely, not included in the compute price shown here. Treat the sizing match as a starting point for comparison, not a guarantee of equivalent real-world throughput, and benchmark your actual workload before committing to a migration based on spec sheets alone.