Moving a workload, or pricing a quote from another cloud? Enter the instance you run today and see the closest equivalent on the other clouds, with the monthly list price of each. Prices come from the vendors' own price lists (read 2026-11-03).
Try: , , , ,
Loading the price lists…
Example: what is m7g.large on the other clouds?
Closest equivalents to m7g.large (2 vCPU, 8 GB, arm64) in India (Mumbai / Pune). Month = hourly price × 730 hours, on-demand Linux.
Same shape. The same number of vCPUs, and at least as much memory (up to 1.5 times as much, shown as "more memory than asked"), on the same processor architecture.
Fixed-performance only. Burstable sizes (AWS T series, Azure B series) are never matched against fixed-performance ones, and GPU instances are left out.
AWS and Azure sell fixed sizes, so each match is a real size. Up to three are shown, cheapest first. Older generations can be cheaper and slower, so check the generation before choosing.
Google Cloud and Oracle bill the cores and memory you pick, so the price is cores × core rate + GB × memory rate for each machine family. Oracle sells x86 cores in pairs (one OCPU is two vCPUs), so an odd vCPU count is rounded up.
Gaps are shown, not zeroed. Azure covers only the Dsv5, Dlsv5, Dpsv5, Esv5 and Fsv2 series so far. A cloud with no match shows why, and no cloud is called lowest while one is missing.
Prices are on-demand Linux list prices before discounts, commitments and tax. A month is the hourly price times 730 hours. Network, storage and licence differences are not part of the comparison; use the estimators for a whole workload, or browse every size in the virtual machine catalog. Methodology.
CloudCostCompare