Speed
1.2×
faster than GitHub-hosted (ubuntu-latest)
7th of 9 by speed
Cost per job
−62%
vs GitHub-hosted (ubuntu-latest)
at 100,000 jobs a month; 60% lower at 10,000, 49% lower at 1,000
Lowest median time
0 / 12
workloads
none of the workloads
Failed runs
0 of 360
no failed measured run
across all workloads
Speed on each workload
× faster than GitHub-hosted (ubuntu-latest) on the same workloadConfiguration
Self-hosted · 2 vCPU / 8 GiB- Measured configuration
- Self-hosted: AWS EC2 m8a.large, one VM per job (on-demand)
- Size
- 2 vCPU / 8 GiB
- Label or setup
self-hosted label of the deployment- Price as measured
- $0.0020287 a minute, each job rounded up to a whole minute; list price on 2026-10-05 (source)
- Notes
- All-in (METHOD.md, Cost): instance seconds, root volume, public IPv4 and internet egress measured per job (cost_ec2.py, VPC flow logs), plus the launcher host as a fixed monthly cost.
- Definition
- providers/ec2-eph.json
Per workload
Green marks the lowest median time on the workload. The burst row covers all 20 jobs.| Workload | Median time | vs GitHub-hosted (ubuntu-latest) | p95 time | Mean cost | vs GitHub-hosted (ubuntu-latest) | Successful runs | Rank |
|---|---|---|---|---|---|---|---|
| Burst: 20 jobs at once | 161 s | 1.6x faster | 182 s | $0.111 | 78% lower | 30 of 30 | 6th |
| cli | 357 s | 1.0x faster | 398 s | $0.014 | 65% lower | 30 of 30 | 7th |
| codex-lint | 512 s | 1.7x faster | 598 s | $0.049 | 44% lower | 30 of 30 | 5th |
| container | 247 s | 1.0x slower | 300 s | $0.044 | 64% higher | 30 of 30 | 8th |
| fd | 114 s | 1.2x slower | 139 s | $0.0047 | 61% lower | 30 of 30 | 9th |
| flask | 70 s | 3.0x slower | 134 s | $0.0036 | 40% lower | 30 of 30 | 9th |
| guava | 717 s | 1.6x faster | 783 s | $0.032 | 71% lower | 30 of 30 | 3rd |
| monorepo | 281 s | 1.4x faster | 342 s | $0.012 | 71% lower | 30 of 30 | 7th |
| pydantic | 154 s | 1.2x faster | 184 s | $0.0064 | 68% lower | 30 of 30 | 7th |
| ts | 433 s | 1.8x faster | 467 s | $0.017 | 78% lower | 30 of 30 | 4th |
| vite | 403 s | 1.5x faster | 503 s | $0.016 | 73% lower | 30 of 30 | 6th |
| vue | 117 s | 1.0x faster | 155 s | $0.0051 | 63% lower | 30 of 30 | 8th |
- Slower than GitHub-hosted (ubuntu-latest)
- container (1.0x slower), fd (1.2x slower), flask (3.0x slower)
- Burst queue, p95
- 93 s, against 31 s for GitHub-hosted (ubuntu-latest)
Indexes and 95% intervals
- Time index
- 0.869 (95% interval 0.833 to 0.947)
- Cost index
- 0.383 (0.358 to 0.405) at 100,000 jobs a month
Compare with…
Head-to-head pages: speed, cost, failures and each workloadVitko Runners vs Self-hosted runners on AWSBlacksmith vs Self-hosted runners on AWSUbicloud vs Self-hosted runners on AWSRunsOn vs Self-hosted runners on AWSSelf-hosted runners on AWS vs Self-hosted runners on Google CloudSelf-hosted runners on AWS vs GitHub-hosted larger runner (4 vCPU)Self-hosted runners on AWS vs GitHub-hosted (ubuntu-latest)