EC2 instance types at 2 vCPU, scored & priced.

Every type is launched on its 2 vCPU size and measured by the open benchmark harness: PassMark's CPU Single Threaded and CPU Mark, sysbench on one thread and on every thread, and 7-Zip's built-in benchmark. AWS often runs custom CPU SKUs you won't find in PassMark's public lists, so every number here comes from a live instance.

Each type's numbers come from one suite, the one that measured it in the most runs: the monthly EC2 CPU sweep, or the hardware suite for the 12 types RunsOn's runner benchmarks also run every few days, so this tab and the CPU page show the same figure. Values are medians over the runs of the current workload definition: the sweep's last 75 days, 30 for the other suites.

A type takes a place (the # column, a leader, a ranked card) once it has 3 runs; below that it is listed with its run count and sorts like any other row. Places exist on the scores and the per-dollar figures, not on a price, a band or a name. A type within 5% of the type heading its group shares that place (“=02”), so near-identical scores are never ranked apart; a leader's near ties are named with it, and a lead under 5%, or with a faster type still under 3 runs, reads “fastest measured”.

Spot is the 7-day average of the cheapest zone's spot price in each region: the RunsOn finder records every zone every 90 minutes, and the average of the cheapest at each fetch is what an hour costs, without one day's swing or one historical low. On-demand is the list price, and savings compare spot with it. Interrupt is AWS Spot Advisor's band. CPU Mark per $ and single-thread per $ divide the score by the spot price of an hour of the instance alone (the CPU page adds the EBS volume); burstable t types are left out of both, as their score is a burst on CPU credits. RunsOn launches any of these types ephemerally, in your own AWS account. More in the methodology.