18 Nov, 2026Aishwarya Palta
Which Monk CI Runner Size Fits Your Workload?
Choose a Monk CI runner from 2 to 32 vCPUs. Compare memory, workloads and costs, then measure whether a larger runner improves your GitHub Actions jobs.
Which Monk CI Runner Size Fits Your Workload?
Choose your Monk CI runner size by the job's CPU use, peak memory and cost per completed build. Monk CI provides 2–32 vCPUs for GitHub Actions jobs, so linting and parallel compilation can use different sizes in the same workflow.
Compare Monk CI runner sizes
These are starting points for testing, not measured performance guarantees. A small repository can have expensive tests; a large repository can have a lightweight lint job.
| vCPU | RAM | Workload to try | Additional usage / minute |
| 2 | 7 GB | Linting, small unit suites, deployment scripts | $0.004 |
| 4 | 15 GB | Frontend builds, backend unit tests | $0.008 |
| 8 | 31 GB | Parallel monorepo tasks, container integration tests | $0.016 |
| 16 | 62 GB | Parallel compilation, memory-heavy test workers | $0.032 |
| 32 | 124 GB | Highly parallel builds and large test suites | $0.064 |
Rates use the published $0.002 per vCPU-minute for additional usage; subscriptions and other charges are separate. See runner specifications and pricing.
Match the size to the bottleneck
Inspect individual step durations before upgrading the entire job:
- Many CPU cores stay busy : try the next size and check your tool's worker limits.
- Memory runs out: add RAM or reduce concurrent workers.
- One core stays busy: investigate serial work before adding cores.
- Downloads or service waits dominate: inspect cache hits, transfers and readiness checks. See the [Actions cache guide] for cache diagnosis.
Choose per job. A workflow can use 2 vCPUs for linting and 16 for compilation without allocating 16 to both.
Measure speed and cost together
Compare adjacent sizes on the same commit with identical dependencies and test coverage. Run each several times, alternating sizes; record worker settings and separate cold-cache from warm-cache results.
Compare median runtime and variation, peak memory, failures and total compute cost. Track queueing separately so you also understand the developer's full wait.
For additional usage:
Estimated compute cost = billed minutes × vCPUs × $0.002
Why Monk CI is better fit
Compared with standard GitHub-hosted Linux x64 runners, Monk CI offers explicit choices up to 32 vCPUs while retaining GitHub Actions workflow syntax. GitHub's standard Ubuntu VMs currently provide 2 CPUs/8 GB for private repositories and 4 CPUs/16 GB for public repositories. [GitHub specifications].
Monk CI provides High core performance with cost optimization
| Repo | Github Hosted Runners | Monk CI Runners | Speed Increase | saved per build |
| Java (Kafka) | 1328s | 162s | 8.2x | 19m 26s |
| Rust(Helix) | 1506s | 183s | 8.2x | 22m 03s |
| Go(Traefik) | 1945s | 293s | 6.6x | 27m 32s |
| Rust(Zed) | 3473s | 724s | 4.8x | 45m 49s |