Every EC2 instance, scored & priced.
Single-thread CPU, real on-demand & spot pricing, and observed spot-interruption frequency for every instance type worth running CI on — across four regions. Sort it, filter it, and open any type for the per-region detail. RunsOn launches all of these in your own AWS account.
Looking for a specific shape? The EC2 instance finder filters and compares by CPU, price, memory and architecture in seconds.
ec2 instance finder ↗| # | instance | processor | arch | cpu speed | on-demand | spot / hr | savings | interrupt |
|---|---|---|---|---|---|---|---|---|
| 01 | m8azn.large→ family m8azn | AMD AMD EPYC 9575F | x64 | 4,262 | $0.2065 | $0.0648/hr | 66% | <5% |
| 02 | m8a.large→ family m8a | AMD AMD EPYC 9R45 | x64 | 3,811 | $0.1217 | $0.0419/hr | 63% | >20% |
| 03 | t8i.small→ family t8i | Intel Intel Xeon Scalable (Granite Rapids) | x64 | 3,193 | $0.0499 | $0.0215/hr | — | n/a |
| 04 | r8i.large→ family r8i | Intel Intel Xeon Scalable (Granite Rapids) | x64 | 3,190 | $0.1389 | $0.0415/hr | 63% | >20% |
| 05 | m8i.large→ family m8i | Intel Intel Xeon Scalable (Granite Rapids) | x64 | 3,189 | $0.1058 | $0.0413/hr | 58% | 10-15% |
| 06 | m8i-flex.large→ family m8i-flex | Intel Intel Xeon Scalable (Granite Rapids) | x64 | 3,189 | $0.1006 | $0.0419/hr | 23% | <5% |
| 07 | r8i-flex.large→ family r8i-flex | Intel Intel Xeon Scalable (Granite Rapids) | x64 | 3,181 | $0.1320 | $0.0535/hr | 56% | >20% |
| 08 | r7iz.large→ family r7iz | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 3,062 | $0.1860 | $0.0525/hr | 69% | 5-10% |
| 09 | i7i.large→ family i7i | Intel Intel Xeon Scalable (Emerald Rapids) | x64 | 2,977 | $0.1888 | $0.0488/hr | 71% | <5% |
| 10 | i7ie.large→ family i7ie | Intel Intel Xeon Scalable (Emerald Rapids) | x64 | 2,970 | $0.2599 | $0.0604/hr | 72% | <5% |
| 11 | m7i-flex.large→ family m7i-flex | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 2,881 | $0.0958 | $0.0407/hr | 58% | 10-15% |
| 12 | c7a.large→ family c7a | AMD AMD EPYC 9R14 | x64 | 2,875 | $0.1026 | $0.0312/hr | 65% | 5-10% |
| 13 | m7a.large→ family m7a | AMD AMD EPYC 9R14 | x64 | 2,874 | $0.1159 | $0.0455/hr | 61% | >20% |
| 14 | r7a.large→ family r7a | AMD AMD EPYC 9R14 | x64 | 2,873 | $0.1522 | $0.0676/hr | 61% | 15-20% |
| 15 | m7i.large→ family m7i | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 2,834 | $0.1008 | $0.0462/hr | 61% | 15-20% |
| 16 | c7i.large→ family c7i | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 2,776 | $0.0892 | $0.0322/hr | 61% | >20% |
| 17 | r7i.large→ family r7i | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 2,774 | $0.1323 | $0.0516/hr | 65% | >20% |
| 18 | c7i-flex.large→ family c7i-flex | Intel Intel Xeon Scalable (Sapphire Rapids) | x64 | 2,694 | $0.0848 | $0.0303/hr | 63% | 5-10% |
| 19 | c6a.large→ family c6a | AMD AMD EPYC 7R13 | x64 | 2,575 | $0.0765 | $0.0303/hr | 59% | 5-10% |
| 20 | m6a.large→ family m6a | AMD AMD EPYC 7R13 | x64 | 2,572 | $0.0864 | $0.0389/hr | 54% | 10-15% |
| 21 | m5zn.large→ family m5zn | Intel Intel Xeon Platinum 8252 | x64 | 2,487 | $0.1652 | $0.0522/hr | 66% | 5-10% |
| 22 | z1d.large→ family z1d | Intel Intel Xeon Platinum 8151 | x64 | 2,303 | $0.1860 | $0.0459/hr | 73% | >20% |
| 23 | c6id.large→ family c6id | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,266 | $0.1008 | $0.0363/hr | 62% | >20% |
| 24 | c6i.large→ family c6i | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,264 | $0.0850 | $0.0381/hr | 59% | 15-20% |
| 25 | r6i.large→ family r6i | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,263 | $0.1260 | $0.0520/hr | 63% | 10-15% |
| 26 | r6idn.large→ family r6idn | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,263 | $0.1954 | $0.0583/hr | 66% | 10-15% |
| 27 | m6i.large→ family m6i | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,262 | $0.0960 | $0.0404/hr | 61% | >20% |
| 28 | m6id.large→ family m6id | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,262 | $0.1187 | $0.0446/hr | 62% | 10-15% |
| 29 | r6id.large→ family r6id | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,260 | $0.1512 | $0.0547/hr | 65% | 15-20% |
| 30 | m6idn.large→ family m6idn | Intel Intel Xeon 8375C (Ice Lake) | x64 | 2,258 | $0.1591 | $0.0565/hr | 60% | <5% |
| 31 | t3.medium→ family t3 | Intel Intel Skylake E5 2686 v5 | x64 | 1,694 | $0.0416 | $0.0166/hr | 60% | >20% |
| 32 | m4.large→ family m4 | Intel Intel Xeon E5-2676 v3 (Haswell) | x64 | 1,545 | $0.1000 | $0.0532/hr | 52% | >20% |
| 33 | m5a.large→ family m5a | AMD AMD EPYC 7571 | x64 | 1,398 | $0.0860 | $0.0392/hr | 55% | 10-15% |
| 34 | t3a.medium→ family t3a | AMD AMD EPYC 7571 | x64 | 1,364 | $0.0376 | $0.0174/hr | 53% | >20% |
What the data says
Single-thread CPU leaders by architecture. The newest families — AMD EPYC 9-series on x64, Graviton4 (Neoverse-V2) on arm64 — set the pace.
About these benchmarks
AWS sometimes runs custom CPU SKUs you won't find on Passmark — so it's genuinely hard to know which instance is fastest. Every type here is launched on its 2 vCPU variant and measured with the Passmark single-thread metric, averaged over many days.
Pricing is real: on-demand list price and the trailing spot average per region, with the resulting savings. Interrupt frequency is AWS's published spot-advisor band. As always, run your own benchmarks — this points you in the right direction.
The benchmark is refreshed regularly so RunsOn users can pick the best-performing EC2 instance for their GitHub Actions workflows — launched ephemerally, in your own account.
Every instance we track
The leaderboard shows one architecture at a time. Every tracked type has its own page with per-region on-demand and spot pricing, savings, and the AWS spot-advisor interruption band.
x64 34 types
- c6a.large
- c6i.large
- c6id.large
- c7a.large
- c7i-flex.large
- c7i.large
- i7i.large
- i7ie.large
- m4.large
- m5a.large
- m5zn.large
- m6a.large
- m6i.large
- m6id.large
- m6idn.large
- m7a.large
- m7i-flex.large
- m7i.large
- m8a.large
- m8azn.large
- m8i-flex.large
- m8i.large
- r6i.large
- r6id.large
- r6idn.large
- r7a.large
- r7i.large
- r7iz.large
- r8i-flex.large
- r8i.large
- t3.medium
- t3a.medium
- t8i.small
- z1d.large
Any instance, in your own account.
The label is a query — name the family or the CPU count and RunsOn launches it ephemerally on spot, one CloudFormation stack, ~10 minutes.