v3.5 changelog
quickstart →
Pricing Benchmarks Docs Changelog Blog Troubleshooting GitHub Install in 10 min

GitHub Actions runner benchmarks

At 4 vCPU, RunsOn is the best value on all three builds.

Rust spotify-player

RunsOn c8a.xlarge · EBS gp3

2m 00s $0.0036 per build

GitHub 4-core: 13× the cost, 2.0× the time, 9th of 9

TypeScript opencode

RunsOn r8a.xlarge · tmpfs

1m 01s $0.0033 per build

GitHub 4-core: 7.9× the cost, 2.2× the time, 7th of 8

Docker PostHog image

RunsOn r8a.xlarge · tmpfs

5m 11s $0.0123 per build

GitHub 4-core: 15× the cost, 2.9× the time, 8th of 8

Best value per provider

Each provider's best variantEach provider's best 2 vCPU variantEach provider's best 4 vCPU variantEach provider's best 8 vCPU variant, ranked by time × cost per build.; shades are build phases in order.; build time on the right.

Rustspotify-player
  1. RunsOn4 vCPU · 8 GBc8a.xlarge · EBS gp32m 00s, cost per build spot $0.0036, on-demand $0.0092, best-value of 9 machines; phases: setup 13s, cargo test 1m 01s, clippy 34s, runner overhead 12s, ×1.00
  2. Ubicloud4 vCPU · 16 GBpremium2m 24s, cost per build $0.0096, best-value of 2 machines; phases: setup 20s, cargo test 1m 14s, clippy 40s, runner overhead 10s, ×3.18
  3. Namespace4 vCPU · 8 GB4x81m 59s, cost per build $0.0119, best-value of 2 machines; phases: setup 10s, cargo test 1m 05s, clippy 35s, runner overhead 9s, ×3.25
  4. StarSling4 vCPU · 16 GB1m 49s, cost per build $0.0145; phases: setup 18s, cargo test 53s, clippy 28s, runner overhead 10s, ×3.64
  5. Avrea4 vCPU · 16 GB2m 04s, cost per build $0.0165; phases: setup 12s, cargo test 1m 07s, clippy 34s, runner overhead 11s, ×4.71
  6. Blacksmith4 vCPU · 16 GB2m 12s, cost per build $0.0176; phases: setup 17s, cargo test 1m 06s, clippy 38s, runner overhead 11s, ×5.34
  7. Warpbuild4 vCPU · 16 GB2m 27s, cost per build $0.0196; phases: setup 15s, cargo test 1m 22s, clippy 41s, runner overhead 9s, ×6.62
  8. AWS CodeBuild4 vCPU · 8 GBmedium3m 50s, cost per build $0.0383; phases: setup 33s, cargo test 1m 58s, clippy 1m 05s, runner overhead 14s, ×20
  9. GitHub4 vCPU · 16 GB3m 56s, cost per build $0.0472; phases: setup 21s, cargo test 2m 12s, clippy 1m 13s, runner overhead 10s, ×26

Bar shades, light to dark, in execution order: setup = checkout + packages + toolchaincargo fmtcargo testclippyrunner overhead

cost per build× = time × cost against the best

TypeScriptopencode
  1. RunsOn4 vCPU · 32 GBr8a.xlarge · tmpfs1m 01s, cost per build spot $0.0033, on-demand $0.0081, best-value of 9 machines; phases: toolchain 1s, clone 1s, bun install 8s, typecheck 36s, runner overhead 15s, ×1.00
  2. Ubicloud4 vCPU · 16 GBstandard1m 47s, cost per build $0.0045, best-value of 2 machines; phases: toolchain 3s, clone 6s, bun install 25s, typecheck 57s, runner overhead 16s, ×2.34
  3. StarSling4 vCPU · 16 GB1m 01s, cost per build $0.0081; phases: toolchain 1s, clone 2s, bun install 10s, typecheck 38s, runner overhead 10s, ×2.43
  4. Blacksmith4 vCPU · 16 GB1m 05s, cost per build $0.0087; phases: toolchain 1s, clone 2s, bun install 10s, typecheck 40s, runner overhead 12s, ×2.76
  5. Warpbuild4 vCPU · 16 GB1m 24s, cost per build $0.0112; phases: toolchain 1s, clone 1s, bun install 14s, typecheck 59s, runner overhead 9s, ×4.61
  6. Namespace4 vCPU · 16 GB4x161m 13s, cost per build $0.0146; phases: toolchain 1s, clone 1s, bun install 29s, typecheck 35s, runner overhead 7s, ×5.22
  7. GitHub4 vCPU · 16 GB2m 12s, cost per build $0.0264; phases: toolchain 1s, clone 2s, bun install 23s, typecheck 1m 28s, runner overhead 18s, ×17
  8. Avrea4 vCPU · 16 GB44s, cost per build $0.0058, 2 runs, not ranked; phases: toolchain 1s, clone 3s, bun install 7s, typecheck 22s, runner overhead 11s, 2 runs
  9. AWS CodeBuildfailed · out of memory

Bar shades, light to dark, in execution order: toolchainclonebun installtypecheckrunner overhead

cost per build× = time × cost against the best

Faded bars are provisional: under 3 runs, not ranked, no ×.

DockerPostHog image
  1. RunsOn4 vCPU · 32 GBr8a.xlarge · tmpfs5m 11s, cost per build spot $0.0123, on-demand $0.0308, best-value of 8 machines; phases: fetch source 6s, buildx setup 3s, docker build 4m 40s, runner overhead 22s, ×1.00
  2. Blacksmith4 vCPU · 16 GB5m 47s, cost per build $0.0463; phases: fetch source 21s, buildx setup 2s, docker build 5m 00s, runner overhead 24s, ×4.19
  3. StarSling4 vCPU · 16 GB5m 55s, cost per build $0.0473; phases: fetch source 23s, buildx setup 4s, docker build 5m 04s, runner overhead 24s, ×4.39
  4. Namespace4 vCPU · 16 GB4x165m 27s, cost per build $0.0654; phases: fetch source 22s, buildx setup 3s, docker build 4m 44s, runner overhead 19s, ×5.58
  5. Ubicloud4 vCPU · 16 GBstandard12m 03s, cost per build $0.0301, best-value of 2 machines; phases: fetch source 18s, buildx setup 12s, docker build 11m 02s, runner overhead 31s, ×5.68
  6. Avrea4 vCPU · 16 GB7m 19s, cost per build $0.0585; phases: fetch source 21s, buildx setup 3s, docker build 6m 23s, runner overhead 32s, ×6.71
  7. Warpbuild4 vCPU · 16 GB8m 01s, cost per build $0.0641; phases: fetch source 23s, buildx setup 4s, docker build 7m 15s, runner overhead 19s, ×8.05
  8. GitHub4 vCPU · 16 GB15m 09s, cost per build $0.1818; phases: fetch source 25s, buildx setup 12s, docker build 12m 41s, runner overhead 1m 51s, ×43
  9. AWS CodeBuildnot run: needs more than 8 GB of memory

Bar shades, light to dark, in execution order: fetch sourcebuildx setupdocker buildrunner overhead

cost per build× = time × cost against the best

Each provider's best-value machine (lowest time × cost among its machines within 15% of its fastest of the same size) at 4 vCPU. Rust: RunsOn tested 9, Namespace 2, Ubicloud 2, the others 1; TypeScript: RunsOn tested 9, Ubicloud 2, the others 1; Docker: RunsOn tested 8, Ubicloud 2, the others 1.

Queue and consistency

How long jobs wait in a burst, and how far build times move from run to run.

Queue under a burst15 jobs at once · each provider's quickest-starting runner at the chosen size · ranked from 3 runs
  1. GitHub 2 vCPU · 8 GB p50 4s, p10–p90 3s to 5s, min 2s, max 6s, 75 samples
  2. StarSling 8 vCPU · 32 GB p50 9s, p10–p90 7s to 1m 18s, min 6s, max 1m 35s, 75 samples
  3. Blacksmith 4 vCPU · 16 GB p50 10s, p10–p90 7s to 14s, min 7s, max 17s, 75 samples
  4. Avrea 4 vCPU · 16 GB p50 12s, p10–p90 10s to 14s, min 9s, max 15s, 75 samples
  5. Warpbuild 4 vCPU · 16 GB p50 13s, p10–p90 11s to 14s, min 11s, max 32s, 75 samples
  6. Namespace 4 vCPU · 8 GB 4x8 p50 14s, p10–p90 10s to 20s, min 8s, max 27s, 75 samples
  7. RunsOn 2 vCPU · 16 GB r8a.large · tmpfs 3 of 5 bursts p50 16s, p10–p90 15s to 19s, min 15s, max 21s, 45 samples
  8. Ubicloud 8 vCPU · 32 GB premium p50 18s, p10–p90 17s to 23s, min 15s, max 26s, 75 samples
  9. AWS CodeBuild 8 vCPU · 16 GB large p50 20s, p10–p90 19s to 24s, min 18s, max 34s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. GitHub 2 vCPU · 8 GB p50 4s, p10–p90 3s to 5s, min 2s, max 6s, 75 samples
  2. Blacksmith 2 vCPU · 8 GB p50 11s, p10–p90 7s to 13s, min 7s, max 14s, 75 samples
  3. Namespace 2 vCPU · 8 GB 2x8 p50 14s, p10–p90 8s to 21s, min 7s, max 25s, 75 samples
  4. Warpbuild 2 vCPU · 8 GB p50 14s, p10–p90 12s to 17s, min 11s, max 1m 23s, 75 samples
  5. Avrea 2 vCPU · 8 GB p50 15s, p10–p90 10s to 19s, min 8s, max 20s, 75 samples
  6. RunsOn 2 vCPU · 16 GB r8a.large · tmpfs 3 of 5 bursts p50 16s, p10–p90 15s to 19s, min 15s, max 21s, 45 samples
  7. Ubicloud 2 vCPU · 8 GB standard p50 18s, p10–p90 10s to 28s, min 9s, max 51s, 75 samples
  8. AWS CodeBuild 2 vCPU · 3 GB small p50 21s, p10–p90 19s to 24s, min 17s, max 29s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. GitHub 4 vCPU · 16 GB p50 4s, p10–p90 3s to 6s, min 2s, max 42s, 75 samples
  2. Blacksmith 4 vCPU · 16 GB p50 10s, p10–p90 7s to 14s, min 7s, max 17s, 75 samples
  3. StarSling 4 vCPU · 16 GB p50 10s, p10–p90 8s to 13s, min 7s, max 14s, 75 samples
  4. Avrea 4 vCPU · 16 GB p50 12s, p10–p90 10s to 14s, min 9s, max 15s, 75 samples
  5. Warpbuild 4 vCPU · 16 GB p50 13s, p10–p90 11s to 14s, min 11s, max 32s, 75 samples
  6. Namespace 4 vCPU · 8 GB 4x8 p50 14s, p10–p90 10s to 20s, min 8s, max 27s, 75 samples
  7. RunsOn 4 vCPU · 32 GB r8a.xlarge · tmpfs p50 17s, p10–p90 16s to 21s, min 15s, max 22s, 75 samples
  8. Ubicloud 4 vCPU · 16 GB premium p50 20s, p10–p90 17s to 28s, min 10s, max 31s, 75 samples
  9. AWS CodeBuild 4 vCPU · 8 GB medium p50 21s, p10–p90 19s to 24s, min 18s, max 26s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. StarSling 8 vCPU · 32 GB p50 9s, p10–p90 7s to 1m 18s, min 6s, max 1m 35s, 75 samples
  2. Blacksmith 8 vCPU · 32 GB p50 11s, p10–p90 8s to 13s, min 8s, max 15s, 75 samples
  3. Namespace 8 vCPU · 32 GB 8x32 p50 14s, p10–p90 8s to 23s, min 7s, max 1m 29s, 75 samples
  4. Warpbuild 8 vCPU · 32 GB p50 15s, p10–p90 10s to 20s, min 7s, max 23s, 75 samples
  5. RunsOn 8 vCPU · 32 GB m8i-flex.2xlarge · EBS gp3 p50 17s, p10–p90 15s to 21s, min 15s, max 24s, 75 samples
  6. Ubicloud 8 vCPU · 32 GB premium p50 18s, p10–p90 17s to 23s, min 15s, max 26s, 75 samples
  7. AWS CodeBuild 8 vCPU · 16 GB large p50 20s, p10–p90 19s to 24s, min 18s, max 34s, 75 samples
  8. GitHub 8 vCPU · 32 GB p50 1m 16s, p10–p90 1m 00s to 1m 34s, min 3s, max 2m 12s, 75 samples
  9. Avrea 8 vCPU · 32 GB 12 at a time: plan limit p50 12s, min 8s, max 1m 34s, 75 samples, not ranked: 12 at a time: plan limit
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. GitHub 4 vCPU · 16 GB p50 6s, p10–p90 5s to 8s, min 5s, max 9s, 75 samples
  2. Ubicloud 2 vCPU · 8 GB standard p50 11s, p10–p90 8s to 35s, min 8s, max 1m 01s, 75 samples
  3. Namespace 2 vCPU · 8 GB Apple silicon 2x8 3 of 5 bursts p50 13s, p10–p90 9s to 19s, min 9s, max 21s, 45 samples
  4. Warpbuild 6 vCPU · 14 GB xfast (Apple M4 Pro) p50 14s, p10–p90 12s to 43s, min 12s, max 7m 36s, 75 samples
  5. Blacksmith 8 vCPU · 32 GB p50 15s, p10–p90 12s to 16s, min 12s, max 21s, 75 samples
  6. RunsOn 2 vCPU · 16 GB r9g.large · tmpfs 3 of 5 bursts p50 17s, p10–p90 15s to 21s, min 14s, max 23s, 45 samples
  7. Avrea 4 vCPU · 16 GB p50 23s, p10–p90 22s to 33s, min 20s, max 1m 26s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. GitHub 2 vCPU · 8 GB p50 7s, p10–p90 5s to 8s, min 5s, max 9s, 75 samples
  2. Ubicloud 2 vCPU · 8 GB standard p50 11s, p10–p90 8s to 35s, min 8s, max 1m 01s, 75 samples
  3. Namespace 2 vCPU · 8 GB Apple silicon 2x8 3 of 5 bursts p50 13s, p10–p90 9s to 19s, min 9s, max 21s, 45 samples
  4. Blacksmith 2 vCPU · 6 GB p50 16s, p10–p90 13s to 19s, min 11s, max 28s, 75 samples
  5. RunsOn 2 vCPU · 16 GB r9g.large · tmpfs 3 of 5 bursts p50 17s, p10–p90 15s to 21s, min 14s, max 23s, 45 samples
  6. Avrea 2 vCPU · 8 GB p50 24s, p10–p90 22s to 28s, min 21s, max 33s, 75 samples
  7. Warpbuild 2 vCPU · 8 GB p50 38s, p10–p90 35s to 46s, min 8s, max 56s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. GitHub 4 vCPU · 16 GB p50 6s, p10–p90 5s to 8s, min 5s, max 9s, 75 samples
  2. Namespace 4 vCPU · 16 GB Apple silicon 4x16 p50 14s, p10–p90 11s to 20s, min 2s, max 25s, 75 samples
  3. Blacksmith 4 vCPU · 16 GB p50 15s, p10–p90 12s to 17s, min 10s, max 26s, 75 samples
  4. RunsOn 4 vCPU · 32 GB r9g.xlarge · tmpfs 4 of 5 bursts p50 18s, p10–p90 15s to 20s, min 14s, max 21s, 60 samples
  5. Avrea 4 vCPU · 16 GB p50 23s, p10–p90 22s to 33s, min 20s, max 1m 26s, 75 samples
  6. Ubicloud 4 vCPU · 16 GB standard p50 23s, p10–p90 10s to 32s, min 9s, max 44s, 75 samples
  7. Warpbuild 4 vCPU · 16 GB p50 37s, p10–p90 18s to 1m 19s, min 6s, max 1m 35s, 75 samples
one job p50 p10–p90 min–max wait before the job starts, log scale
  1. Blacksmith 8 vCPU · 32 GB p50 15s, p10–p90 12s to 16s, min 12s, max 21s, 75 samples
  2. RunsOn 8 vCPU · 32 GB m9g.2xlarge · tmpfs p50 17s, p10–p90 15s to 21s, min 14s, max 25s, 75 samples
  3. Namespace 8 vCPU · 16 GB 8x16 p50 20s, p10–p90 15s to 26s, min 14s, max 4m 20s, 75 samples
  4. Ubicloud 8 vCPU · 32 GB standard p50 23s, p10–p90 20s to 2m 39s, min 9s, max 4m 19s, 75 samples
  5. Warpbuild 8 vCPU · 32 GB p50 35s, p10–p90 16s to 46s, min 8s, max 49s, 75 samples
  6. Avrea 8 vCPU · 32 GB 12 at a time: plan limit p50 25s, min 19s, max 2m 03s, 75 samples, not ranked: 12 at a time: plan limit
one job p50 p10–p90 min–max wait before the job starts, log scale
Build time, run to runeach provider's machine from the cards above, at the chosen size, with 3+ workflow runs · build steps time, one dot per run · % off its own p50
  1. RunsOn 4 vCPU · 8 GB c8a.xlarge · EBS gp3 spread 4.7% (p10–p90) around a p50 of 1m 51s, min 1m 46s, max 1m 52s, 5 samples
  2. StarSling 4 vCPU · 16 GB spread 11.5% (p10–p90) around a p50 of 1m 43s, min 1m 28s, max 1m 45s, 5 samples
  3. Blacksmith 8 vCPU · 32 GB spread 18.2% (p10–p90) around a p50 of 1m 16s, min 1m 11s, max 1m 26s, 5 samples
  4. Namespace 4 vCPU · 8 GB 4x8 spread 22% (p10–p90) around a p50 of 1m 52s, min 1m 38s, max 2m 05s, 5 samples
  5. Warpbuild 8 vCPU · 32 GB spread 22.9% (p10–p90) around a p50 of 1m 29s, min 1m 17s, max 1m 47s, 5 samples
  6. Avrea 4 vCPU · 16 GB spread 23.6% (p10–p90) around a p50 of 1m 56s, min 1m 51s, max 2m 30s, 5 samples
  7. GitHub 8 vCPU · 32 GB spread 27.1% (p10–p90) around a p50 of 2m 24s, min 1m 48s, max 2m 31s, 5 samples
  8. AWS CodeBuild 8 vCPU · 16 GB large spread 28.9% (p10–p90) around a p50 of 2m 19s, min 2m 02s, max 3m 05s, 5 samples
  9. Ubicloud 2 vCPU · 8 GB standard spread 81% (p10–p90) around a p50 of 3m 59s, min 3m 54s, max 7m 36s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 2 vCPU · 8 GB m8a.large · EBS gp3 p50 2m 49s, min 2m 45s, max 2m 53s, 4 samples, not ranked
  2. Warpbuild 2 vCPU · 8 GB spread 11.2% (p10–p90) around a p50 of 3m 35s, min 3m 32s, max 4m 06s, 5 samples
  3. GitHub 2 vCPU · 8 GB spread 18.4% (p10–p90) around a p50 of 6m 56s, min 5m 26s, max 7m 13s, 5 samples
  4. Avrea 2 vCPU · 8 GB spread 19% (p10–p90) around a p50 of 3m 09s, min 3m 04s, max 3m 48s, 5 samples
  5. Blacksmith 2 vCPU · 8 GB spread 20% (p10–p90) around a p50 of 3m 12s, min 3m 01s, max 3m 57s, 5 samples
  6. Namespace 2 vCPU · 8 GB 2x8 spread 23.4% (p10–p90) around a p50 of 2m 39s, min 2m 36s, max 3m 34s, 5 samples
  7. AWS CodeBuild 2 vCPU · 3 GB small spread 29.4% (p10–p90) around a p50 of 6m 27s, min 6m 08s, max 9m 11s, 5 samples
  8. Ubicloud 2 vCPU · 8 GB standard spread 81% (p10–p90) around a p50 of 3m 59s, min 3m 54s, max 7m 36s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 4 vCPU · 8 GB c8a.xlarge · EBS gp3 spread 4.7% (p10–p90) around a p50 of 1m 51s, min 1m 46s, max 1m 52s, 5 samples
  2. GitHub 4 vCPU · 16 GB spread 7% (p10–p90) around a p50 of 3m 47s, min 3m 35s, max 3m 58s, 5 samples
  3. StarSling 4 vCPU · 16 GB spread 11.5% (p10–p90) around a p50 of 1m 43s, min 1m 28s, max 1m 45s, 5 samples
  4. Warpbuild 4 vCPU · 16 GB spread 13% (p10–p90) around a p50 of 2m 20s, min 1m 59s, max 2m 22s, 5 samples
  5. Blacksmith 4 vCPU · 16 GB spread 15.9% (p10–p90) around a p50 of 2m 01s, min 1m 50s, max 2m 12s, 5 samples
  6. Namespace 4 vCPU · 8 GB 4x8 spread 22% (p10–p90) around a p50 of 1m 52s, min 1m 38s, max 2m 05s, 5 samples
  7. Avrea 4 vCPU · 16 GB spread 23.6% (p10–p90) around a p50 of 1m 56s, min 1m 51s, max 2m 30s, 5 samples
  8. AWS CodeBuild 4 vCPU · 8 GB medium spread 26% (p10–p90) around a p50 of 3m 35s, min 3m 11s, max 4m 24s, 5 samples
  9. Ubicloud 4 vCPU · 16 GB premium spread 87.8% (p10–p90) around a p50 of 2m 13s, min 1m 49s, max 4m 07s, 5 samples
one run p50 p10–p90 min–max
  1. Ubicloud 8 vCPU · 32 GB standard spread 11.7% (p10–p90) around a p50 of 1m 56s, min 1m 52s, max 2m 08s, 5 samples
  2. Avrea 8 vCPU · 32 GB spread 12.3% (p10–p90) around a p50 of 1m 18s, min 1m 13s, max 1m 25s, 5 samples
  3. StarSling 8 vCPU · 32 GB spread 13.1% (p10–p90) around a p50 of 1m 12s, min 1m 01s, max 1m 16s, 5 samples
  4. RunsOn 8 vCPU · 16 GB c8a.2xlarge · EBS gp3 spread 13.5% (p10–p90) around a p50 of 1m 14s, min 1m 13s, max 1m 29s, 5 samples
  5. Blacksmith 8 vCPU · 32 GB spread 18.2% (p10–p90) around a p50 of 1m 16s, min 1m 11s, max 1m 26s, 5 samples
  6. Warpbuild 8 vCPU · 32 GB spread 22.9% (p10–p90) around a p50 of 1m 29s, min 1m 17s, max 1m 47s, 5 samples
  7. Namespace 8 vCPU · 16 GB 8x16 spread 23.3% (p10–p90) around a p50 of 1m 20s, min 1m 08s, max 1m 31s, 5 samples
  8. GitHub 8 vCPU · 32 GB spread 27.1% (p10–p90) around a p50 of 2m 24s, min 1m 48s, max 2m 31s, 5 samples
  9. AWS CodeBuild 8 vCPU · 16 GB large spread 28.9% (p10–p90) around a p50 of 2m 19s, min 2m 02s, max 3m 05s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 8 vCPU · 32 GB m9gd.2xlarge · local NVMe spread 8.6% (p10–p90) around a p50 of 1m 12s, min 1m 11s, max 1m 18s, 5 samples
  2. Blacksmith 8 vCPU · 32 GB spread 9.3% (p10–p90) around a p50 of 3m 10s, min 3m 02s, max 3m 24s, 5 samples
  3. GitHub 4 vCPU · 16 GB spread 9.8% (p10–p90) around a p50 of 3m 10s, min 2m 53s, max 3m 16s, 5 samples
  4. Avrea 2 vCPU · 8 GB spread 13.7% (p10–p90) around a p50 of 2m 14s, min 1m 56s, max 2m 16s, 5 samples
  5. Warpbuild 8 vCPU · 32 GB spread 13.9% (p10–p90) around a p50 of 1m 59s, min 1m 51s, max 2m 10s, 5 samples
  6. Ubicloud 4 vCPU · 16 GB standard spread 28.3% (p10–p90) around a p50 of 4m 31s, min 4m 12s, max 5m 45s, 5 samples
  7. Namespace 4 vCPU · 16 GB Apple silicon 4x16 spread 37.6% (p10–p90) around a p50 of 1m 25s, min 1m 19s, max 2m 11s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 2 vCPU · 8 GB m9gd.large · local NVMe spread 0.9% (p10–p90) around a p50 of 2m 52s, min 2m 51s, max 2m 53s, 5 samples
  2. Warpbuild 2 vCPU · 8 GB spread 6.4% (p10–p90) around a p50 of 4m 46s, min 4m 33s, max 4m 54s, 5 samples
  3. GitHub 2 vCPU · 8 GB spread 7% (p10–p90) around a p50 of 5m 07s, min 5m 06s, max 5m 28s, 5 samples
  4. Ubicloud 2 vCPU · 8 GB standard spread 7.8% (p10–p90) around a p50 of 7m 26s, min 7m 01s, max 7m 46s, 5 samples
  5. Namespace 2 vCPU · 8 GB Apple silicon 2x8 p50 2m 05s, min 2m 04s, max 2m 21s, 3 samples, not ranked
  6. Blacksmith 2 vCPU · 6 GB spread 11.2% (p10–p90) around a p50 of 7m 44s, min 7m 30s, max 8m 43s, 5 samples
  7. Avrea 2 vCPU · 8 GB spread 13.7% (p10–p90) around a p50 of 2m 14s, min 1m 56s, max 2m 16s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 4 vCPU · 16 GB m9g.xlarge · EBS gp3 spread 3.1% (p10–p90) around a p50 of 1m 48s, min 1m 46s, max 1m 51s, 5 samples
  2. Blacksmith 4 vCPU · 16 GB spread 9.2% (p10–p90) around a p50 of 4m 39s, min 4m 27s, max 4m 58s, 5 samples
  3. GitHub 4 vCPU · 16 GB spread 9.8% (p10–p90) around a p50 of 3m 10s, min 2m 53s, max 3m 16s, 5 samples
  4. Warpbuild 4 vCPU · 16 GB spread 11.3% (p10–p90) around a p50 of 2m 59s, min 2m 41s, max 3m 10s, 5 samples
  5. Avrea 4 vCPU · 16 GB spread 27.7% (p10–p90) around a p50 of 1m 36s, min 1m 20s, max 1m 53s, 5 samples
  6. Ubicloud 4 vCPU · 16 GB standard spread 28.3% (p10–p90) around a p50 of 4m 31s, min 4m 12s, max 5m 45s, 5 samples
  7. Namespace 4 vCPU · 16 GB Apple silicon 4x16 spread 37.6% (p10–p90) around a p50 of 1m 25s, min 1m 19s, max 2m 11s, 5 samples
one run p50 p10–p90 min–max
  1. RunsOn 8 vCPU · 32 GB m9gd.2xlarge · local NVMe spread 8.6% (p10–p90) around a p50 of 1m 12s, min 1m 11s, max 1m 18s, 5 samples
  2. Blacksmith 8 vCPU · 32 GB spread 9.3% (p10–p90) around a p50 of 3m 10s, min 3m 02s, max 3m 24s, 5 samples
  3. Namespace 8 vCPU · 16 GB 8x16 spread 11.7% (p10–p90) around a p50 of 3m 55s, min 3m 28s, max 4m 04s, 5 samples
  4. Warpbuild 8 vCPU · 32 GB spread 13.9% (p10–p90) around a p50 of 1m 59s, min 1m 51s, max 2m 10s, 5 samples
  5. Ubicloud 8 vCPU · 32 GB standard spread 21.5% (p10–p90) around a p50 of 3m 29s, min 3m 28s, max 4m 23s, 5 samples
  6. Avrea 8 vCPU · 32 GB spread 23.4% (p10–p90) around a p50 of 1m 17s, min 1m 01s, max 1m 19s, 5 samples
one run p50 p10–p90 min–max
  1. Warpbuild 8 vCPU · 32 GB spread 3.6% (p10–p90) around a p50 of 44s, min 43s, max 45s, 5 samples
  2. GitHub 8 vCPU · 32 GB spread 23.9% (p10–p90) around a p50 of 1m 11s, min 56s, max 1m 15s, 5 samples
  3. Blacksmith 4 vCPU · 16 GB spread 24.4% (p10–p90) around a p50 of 54s, min 48s, max 1m 04s, 5 samples
  4. Ubicloud 8 vCPU · 32 GB premium spread 32% (p10–p90) around a p50 of 40s, min 37s, max 51s, 5 samples
  5. StarSling 4 vCPU · 16 GB spread 39.6% (p10–p90) around a p50 of 52s, min 44s, max 1m 17s, 5 samples
  6. Namespace 8 vCPU · 16 GB 8x16 spread 45.6% (p10–p90) around a p50 of 39s, min 31s, max 54s, 5 samples
  7. AWS CodeBuild 8 vCPU · 16 GB large spread 48.1% (p10–p90) around a p50 of 1m 07s, min 1m 02s, max 1m 45s, 5 samples
  8. RunsOn 8 vCPU · 32 GB m8a.2xlarge · tmpfs spread 101.4% (p10–p90) around a p50 of 29s, min 29s, max 1m 18s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. Blacksmith 2 vCPU · 8 GB spread 12.5% (p10–p90) around a p50 of 1m 46s, min 1m 37s, max 1m 53s, 5 samples
  2. GitHub 2 vCPU · 8 GB p50 3m 44s, min 3m 32s, max 4m 33s, 3 samples, not ranked
  3. RunsOn 2 vCPU · 16 GB r8a.large · tmpfs p50 1m 25s, min 1m 25s, max 2m 11s, 3 samples, not ranked
  4. Warpbuild 2 vCPU · 8 GB spread 48.1% (p10–p90) around a p50 of 2m 38s, min 1m 54s, max 3m 30s, 5 samples
  5. Ubicloud 2 vCPU · 8 GB premium spread 61% (p10–p90) around a p50 of 2m 58s, min 2m 13s, max 4m 36s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. Warpbuild 4 vCPU · 16 GB spread 10.7% (p10–p90) around a p50 of 1m 15s, min 1m 12s, max 1m 24s, 5 samples
  2. GitHub 4 vCPU · 16 GB spread 13.6% (p10–p90) around a p50 of 1m 56s, min 1m 43s, max 2m 06s, 5 samples
  3. Ubicloud 4 vCPU · 16 GB standard spread 20.2% (p10–p90) around a p50 of 1m 30s, min 1m 25s, max 1m 54s, 5 samples
  4. Blacksmith 4 vCPU · 16 GB spread 24.4% (p10–p90) around a p50 of 54s, min 48s, max 1m 04s, 5 samples
  5. StarSling 4 vCPU · 16 GB spread 39.6% (p10–p90) around a p50 of 52s, min 44s, max 1m 17s, 5 samples
  6. Namespace 4 vCPU · 16 GB 4x16 spread 52% (p10–p90) around a p50 of 1m 05s, min 47s, max 1m 34s, 5 samples
  7. RunsOn 4 vCPU · 32 GB r8a.xlarge · tmpfs spread 75.2% (p10–p90) around a p50 of 46s, min 45s, max 1m 40s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. Warpbuild 8 vCPU · 32 GB spread 3.6% (p10–p90) around a p50 of 44s, min 43s, max 45s, 5 samples
  2. Blacksmith 8 vCPU · 32 GB spread 10.3% (p10–p90) around a p50 of 37s, min 33s, max 38s, 5 samples
  3. GitHub 8 vCPU · 32 GB spread 23.9% (p10–p90) around a p50 of 1m 11s, min 56s, max 1m 15s, 5 samples
  4. Ubicloud 8 vCPU · 32 GB premium spread 32% (p10–p90) around a p50 of 40s, min 37s, max 51s, 5 samples
  5. StarSling 8 vCPU · 32 GB spread 42.9% (p10–p90) around a p50 of 42s, min 31s, max 53s, 5 samples
  6. Namespace 8 vCPU · 16 GB 8x16 spread 45.6% (p10–p90) around a p50 of 39s, min 31s, max 54s, 5 samples
  7. AWS CodeBuild 8 vCPU · 16 GB large spread 48.1% (p10–p90) around a p50 of 1m 07s, min 1m 02s, max 1m 45s, 5 samples
  8. RunsOn 8 vCPU · 32 GB m8a.2xlarge · tmpfs spread 101.4% (p10–p90) around a p50 of 29s, min 29s, max 1m 18s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. GitHub 4 vCPU · 16 GB spread 7.1% (p10–p90) around a p50 of 1m 24s, min 1m 24s, max 1m 34s, 5 samples
  2. Blacksmith 8 vCPU · 32 GB spread 19.8% (p10–p90) around a p50 of 1m 20s, min 1m 13s, max 1m 36s, 5 samples
  3. Ubicloud 8 vCPU · 32 GB standard spread 21.8% (p10–p90) around a p50 of 1m 19s, min 1m 11s, max 1m 31s, 5 samples
  4. Namespace 8 vCPU · 16 GB 8x16 spread 27.4% (p10–p90) around a p50 of 1m 43s, min 1m 26s, max 1m 55s, 5 samples
  5. Warpbuild 8 vCPU · 32 GB spread 74% (p10–p90) around a p50 of 43s, min 42s, max 1m 33s, 5 samples
  6. RunsOn 4 vCPU · 16 GB m9g.xlarge · tmpfs spread 420% (p10–p90) around a p50 of 47s, min 44s, max 6m 11s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. GitHub 2 vCPU · 8 GB p50 3m 12s, min 3m 00s, max 3m 14s, 4 samples, not ranked
  2. Ubicloud 2 vCPU · 8 GB standard p50 4m 27s, min 4m 01s, max 4m 38s, 3 samples, not ranked
  3. Blacksmith 2 vCPU · 6 GB spread 11.3% (p10–p90) around a p50 of 6m 52s, min 6m 28s, max 7m 31s, 5 samples
  4. RunsOn 2 vCPU · 8 GB m9gd.large · local NVMe spread 18.4% (p10–p90) around a p50 of 1m 46s, min 1m 39s, max 2m 07s, 6 samples
  5. Warpbuild 2 vCPU · 8 GB spread 70.5% (p10–p90) around a p50 of 6m 07s, min 2m 08s, max 7m 23s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. Blacksmith 4 vCPU · 16 GB spread 7% (p10–p90) around a p50 of 2m 23s, min 2m 14s, max 2m 24s, 5 samples
  2. GitHub 4 vCPU · 16 GB spread 7.1% (p10–p90) around a p50 of 1m 24s, min 1m 24s, max 1m 34s, 5 samples
  3. Ubicloud 4 vCPU · 16 GB standard spread 19.4% (p10–p90) around a p50 of 2m 06s, min 1m 45s, max 2m 17s, 5 samples
  4. Warpbuild 4 vCPU · 16 GB spread 33.6% (p10–p90) around a p50 of 1m 15s, min 1m 09s, max 1m 49s, 5 samples
  5. Namespace 4 vCPU · 16 GB Apple silicon 4x16 spread 207.8% (p10–p90) around a p50 of 46s, min 36s, max 2m 18s, 5 samples
  6. RunsOn 4 vCPU · 16 GB m9g.xlarge · tmpfs spread 420% (p10–p90) around a p50 of 47s, min 44s, max 6m 11s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. Blacksmith 8 vCPU · 32 GB spread 19.8% (p10–p90) around a p50 of 1m 20s, min 1m 13s, max 1m 36s, 5 samples
  2. Ubicloud 8 vCPU · 32 GB standard spread 21.8% (p10–p90) around a p50 of 1m 19s, min 1m 11s, max 1m 31s, 5 samples
  3. Namespace 8 vCPU · 16 GB 8x16 spread 27.4% (p10–p90) around a p50 of 1m 43s, min 1m 26s, max 1m 55s, 5 samples
  4. Warpbuild 8 vCPU · 32 GB spread 74% (p10–p90) around a p50 of 43s, min 42s, max 1m 33s, 5 samples
  5. RunsOn 8 vCPU · 32 GB m9gd.2xlarge · local NVMe spread 89.7% (p10–p90) around a p50 of 31s, min 29s, max 1m 14s, 5 samples
one run p50 p10–p90 min–max

Under 3 runs so far: Avrea.

  1. RunsOn 8 vCPU · 32 GB m8a.2xlarge · tmpfs spread 9.3% (p10–p90) around a p50 of 3m 39s, min 3m 32s, max 3m 58s, 5 samples
  2. Warpbuild 8 vCPU · 32 GB spread 11.8% (p10–p90) around a p50 of 4m 54s, min 4m 34s, max 5m 22s, 7 samples
  3. Namespace 8 vCPU · 16 GB 8x16 spread 12.1% (p10–p90) around a p50 of 3m 56s, min 3m 52s, max 4m 53s, 7 samples
  4. Blacksmith 4 vCPU · 16 GB spread 15.4% (p10–p90) around a p50 of 5m 22s, min 4m 59s, max 5m 57s, 7 samples
  5. Ubicloud 8 vCPU · 32 GB standard spread 22.5% (p10–p90) around a p50 of 7m 50s, min 7m 16s, max 9m 51s, 7 samples
  6. AWS CodeBuild 8 vCPU · 16 GB large spread 23.5% (p10–p90) around a p50 of 12m 03s, min 11m 11s, max 14m 17s, 6 samples
  7. Avrea 8 vCPU · 32 GB spread 32.1% (p10–p90) around a p50 of 4m 47s, min 4m 14s, max 6m 40s, 7 samples
  8. StarSling 4 vCPU · 16 GB spread 36.2% (p10–p90) around a p50 of 5m 28s, min 5m 22s, max 9m 34s, 7 samples
  9. GitHub 8 vCPU · 32 GB spread 273.9% (p10–p90) around a p50 of 7m 18s, min 6m 04s, max 52m 26s, 7 samples
one run p50 p10–p90 min–max
  1. RunsOn 2 vCPU · 16 GB r8a.large · EBS gp3 spread 9.1% (p10–p90) around a p50 of 11m 32s, min 11m 05s, max 12m 13s, 6 samples
one run p50 p10–p90 min–max
  1. RunsOn 4 vCPU · 32 GB r8a.xlarge · tmpfs spread 6.2% (p10–p90) around a p50 of 4m 55s, min 4m 41s, max 5m 06s, 5 samples
  2. Warpbuild 4 vCPU · 16 GB spread 9.4% (p10–p90) around a p50 of 7m 40s, min 6m 54s, max 7m 56s, 7 samples
  3. Namespace 4 vCPU · 16 GB 4x16 spread 11.1% (p10–p90) around a p50 of 5m 08s, min 4m 52s, max 5m 47s, 6 samples
  4. GitHub 4 vCPU · 16 GB spread 14.7% (p10–p90) around a p50 of 13m 13s, min 12m 10s, max 15m 18s, 7 samples
  5. Blacksmith 4 vCPU · 16 GB spread 15.4% (p10–p90) around a p50 of 5m 22s, min 4m 59s, max 5m 57s, 7 samples
  6. Avrea 4 vCPU · 16 GB spread 23% (p10–p90) around a p50 of 6m 47s, min 5m 57s, max 7m 44s, 7 samples
  7. Ubicloud 4 vCPU · 16 GB standard spread 32.3% (p10–p90) around a p50 of 11m 29s, min 10m 02s, max 15m 39s, 7 samples
  8. StarSling 4 vCPU · 16 GB spread 36.2% (p10–p90) around a p50 of 5m 28s, min 5m 22s, max 9m 34s, 7 samples
one run p50 p10–p90 min–max
  1. RunsOn 8 vCPU · 32 GB m8a.2xlarge · tmpfs spread 9.3% (p10–p90) around a p50 of 3m 39s, min 3m 32s, max 3m 58s, 5 samples
  2. StarSling 8 vCPU · 32 GB spread 9.9% (p10–p90) around a p50 of 4m 13s, min 3m 54s, max 4m 31s, 7 samples
  3. Warpbuild 8 vCPU · 32 GB spread 11.8% (p10–p90) around a p50 of 4m 54s, min 4m 34s, max 5m 22s, 7 samples
  4. Namespace 8 vCPU · 16 GB 8x16 spread 12.1% (p10–p90) around a p50 of 3m 56s, min 3m 52s, max 4m 53s, 7 samples
  5. Blacksmith 8 vCPU · 32 GB spread 12.1% (p10–p90) around a p50 of 3m 56s, min 3m 46s, max 4m 29s, 7 samples
  6. Ubicloud 8 vCPU · 32 GB standard spread 22.5% (p10–p90) around a p50 of 7m 50s, min 7m 16s, max 9m 51s, 7 samples
  7. AWS CodeBuild 8 vCPU · 16 GB large spread 23.5% (p10–p90) around a p50 of 12m 03s, min 11m 11s, max 14m 17s, 6 samples
  8. Avrea 8 vCPU · 32 GB spread 32.1% (p10–p90) around a p50 of 4m 47s, min 4m 14s, max 6m 40s, 7 samples
  9. GitHub 8 vCPU · 32 GB spread 273.9% (p10–p90) around a p50 of 7m 18s, min 6m 04s, max 52m 26s, 7 samples
one run p50 p10–p90 min–max
  1. Blacksmith 8 vCPU · 32 GB spread 5.7% (p10–p90) around a p50 of 8m 00s, min 7m 49s, max 8m 39s, 7 samples
  2. Warpbuild 8 vCPU · 32 GB spread 6.6% (p10–p90) around a p50 of 6m 34s, min 6m 12s, max 6m 43s, 7 samples
  3. RunsOn 8 vCPU · 32 GB m9g.2xlarge · tmpfs spread 10.9% (p10–p90) around a p50 of 3m 59s, min 3m 51s, max 4m 23s, 5 samples
  4. Avrea 4 vCPU · 16 GB spread 19.3% (p10–p90) around a p50 of 5m 01s, min 4m 30s, max 6m 01s, 7 samples
  5. Namespace 8 vCPU · 16 GB 8x16 spread 19.4% (p10–p90) around a p50 of 10m 17s, min 9m 32s, max 11m 57s, 7 samples
  6. Ubicloud 4 vCPU · 16 GB standard spread 20.3% (p10–p90) around a p50 of 14m 02s, min 13m 26s, max 17m 09s, 7 samples
  7. GitHub 4 vCPU · 16 GB spread 118.2% (p10–p90) around a p50 of 9m 37s, min 9m 04s, max 35m 34s, 7 samples
one run p50 p10–p90 min–max

No arm64 Docker build finished at 2 vCPU.

  1. RunsOn 4 vCPU · 32 GB r9g.xlarge · tmpfs spread 5.2% (p10–p90) around a p50 of 5m 19s, min 5m 06s, max 5m 27s, 5 samples
  2. Warpbuild 4 vCPU · 16 GB spread 10.5% (p10–p90) around a p50 of 10m 11s, min 9m 09s, max 10m 52s, 7 samples
  3. Namespace 4 vCPU · 16 GB Apple silicon 4x16 spread 13.9% (p10–p90) around a p50 of 4m 31s, min 4m 19s, max 5m 16s, 6 samples
  4. Avrea 4 vCPU · 16 GB spread 19.3% (p10–p90) around a p50 of 5m 01s, min 4m 30s, max 6m 01s, 7 samples
  5. Ubicloud 4 vCPU · 16 GB standard spread 20.3% (p10–p90) around a p50 of 14m 02s, min 13m 26s, max 17m 09s, 7 samples
  6. Blacksmith 4 vCPU · 16 GB spread 41.3% (p10–p90) around a p50 of 12m 40s, min 11m 05s, max 18m 32s, 7 samples
  7. GitHub 4 vCPU · 16 GB spread 118.2% (p10–p90) around a p50 of 9m 37s, min 9m 04s, max 35m 34s, 7 samples
one run p50 p10–p90 min–max
  1. Blacksmith 8 vCPU · 32 GB spread 5.7% (p10–p90) around a p50 of 8m 00s, min 7m 49s, max 8m 39s, 7 samples
  2. Warpbuild 8 vCPU · 32 GB spread 6.6% (p10–p90) around a p50 of 6m 34s, min 6m 12s, max 6m 43s, 7 samples
  3. RunsOn 8 vCPU · 32 GB m9g.2xlarge · tmpfs spread 10.9% (p10–p90) around a p50 of 3m 59s, min 3m 51s, max 4m 23s, 5 samples
  4. Namespace 8 vCPU · 16 GB 8x16 spread 19.4% (p10–p90) around a p50 of 10m 17s, min 9m 32s, max 11m 57s, 7 samples
  5. Ubicloud 8 vCPU · 32 GB standard spread 20.9% (p10–p90) around a p50 of 10m 31s, min 9m 43s, max 12m 19s, 7 samples
  6. Avrea 8 vCPU · 32 GB spread 28.8% (p10–p90) around a p50 of 3m 49s, min 3m 32s, max 4m 44s, 7 samples
one run p50 p10–p90 min–max

Is a bigger runner always worth it?

What doubling the vCPU did to each build, on each provider's own runner line: each step usually saves less time and costs more per build.

Median across providers: from 4 to 8 vCPU, each build ran 22–31% faster for 37–55% more per build; from 2 to 4 vCPU, 36–46% faster, cost per build +8% to +25%.

Median across providers: from 4 to 8 vCPU, each build ran 22–31% faster for 38–55% more per build; from 2 to 4 vCPU, 36–46% faster, cost per build +8% to +27%.

Median across providers: from 4 to 8 vCPU, each build ran 29–32% faster for 36–42% more per build; from 2 to 4 vCPU, 38–54% faster, cost per build −26% to +23%.

Median across providers: from 4 to 8 vCPU, each build ran 29–32% faster for 36–48% more per build; from 2 to 4 vCPU, 38–54% faster, cost per build −26% to +23%.

  • Each line is one provider's runner at 2, 4 and 8 vCPU, its vCPU count in each dot (hollow: under 3 runs so far); RunsOn's is the bold one (m8a / m9g, not its fastest or cheapest).
  • Right costs more per build, lower is faster: the shaded corner, bottom left, is where you want to be.
  • Dashed lines join equal time × cost. A step that drops more steeply than them bought time cheaply; one that runs flatter mostly bought cost.
Rustspotify-player
faster & cheaper$0.0050$0.0100$0.0200$0.05002m 00s5m 00ssame time × costcost per build →build time · faster ↓NamespaceStarSlingRunsOnAvreaBlacksmithWarpbuildUbicloudAWS CodeBuildGitHub

Median across providers: 4 → 8 vCPU: 31% faster, +37% cost (9 providers)
2 → 4 vCPU: 36% faster, +25% cost (8 providers)

faster & cheaper$0.0050$0.0100$0.0200$0.05002m 00s5m 00ssame time × costcost per build →build time · faster ↓NamespaceStarSlingRunsOnAvreaBlacksmithWarpbuildUbicloudAWS CodeBuildGitHub

Median across providers: 4 → 8 vCPU: 31% faster, +38% cost (9 providers)
2 → 4 vCPU: 36% faster, +27% cost (8 providers)

faster & cheaper$0.0050$0.0100$0.0200$0.0500$0.10002m 00s5m 00s10m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildGitHubUbicloudBlacksmithNamespace

Median across providers: 4 → 8 vCPU: 29% faster, +42% cost (6 providers)
2 → 4 vCPU: 38% faster, +23% cost (7 providers)

faster & cheaper$0.0100$0.0200$0.0500$0.10002m 00s5m 00s10m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildGitHubUbicloudBlacksmithNamespace

Median across providers: 4 → 8 vCPU: 29% faster, +48% cost (6 providers)
2 → 4 vCPU: 38% faster, +23% cost (7 providers)

TypeScriptopencode
faster & cheaper$0.0020$0.0050$0.0100$0.020030s1m 00s2m 00ssame time × costcost per build →build time · faster ↓AvreaStarSlingBlacksmithNamespaceRunsOnWarpbuildUbicloudGitHub

Median across providers: 4 → 8 vCPU: 22% faster, +55% cost (7 providers)
2 → 4 vCPU: 46% faster, +8% cost (5 providers)

faster & cheaper$0.0020$0.0050$0.0100$0.020030s1m 00s2m 00ssame time × costcost per build →build time · faster ↓AvreaStarSlingBlacksmithNamespaceRunsOnWarpbuildUbicloudGitHub

Median across providers: 4 → 8 vCPU: 22% faster, +55% cost (7 providers)
2 → 4 vCPU: 46% faster, +8% cost (5 providers)

faster & cheaper$0.0050$0.0100$0.0200$0.050030s1m 00s2m 00s5m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildGitHubUbicloudBlacksmithNamespace

Median across providers: 4 → 8 vCPU: 32% faster, +36% cost (5 providers)
2 → 4 vCPU: 54% faster, −26% cost (5 providers)

faster & cheaper$0.0050$0.0100$0.0200$0.050030s1m 00s2m 00s5m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildGitHubUbicloudBlacksmithNamespace

Median across providers: 4 → 8 vCPU: 32% faster, +36% cost (5 providers)
2 → 4 vCPU: 54% faster, −26% cost (5 providers)

DockerPostHog image
faster & cheaper$0.0200$0.0500$0.1000$0.20005m 00s10m 00s15m 00ssame time × costcost per build →build time · faster ↓NamespaceBlacksmithStarSlingAvreaWarpbuildRunsOnUbicloudGitHub

Median across providers: 4 → 8 vCPU: 30% faster, +40% cost (8 providers)

faster & cheaper$0.0500$0.1000$0.20005m 00s10m 00s15m 00ssame time × costcost per build →build time · faster ↓NamespaceBlacksmithStarSlingAvreaWarpbuildRunsOnUbicloudGitHub

Median across providers: 4 → 8 vCPU: 30% faster, +40% cost (8 providers)

faster & cheaper$0.0200$0.0500$0.1000$0.20005m 00s10m 00s15m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildBlacksmithNamespaceUbicloud

Median across providers: 4 → 8 vCPU: 29% faster, +39% cost (6 providers)

faster & cheaper$0.0500$0.1000$0.20005m 00s10m 00s15m 00ssame time × costcost per build →build time · faster ↓AvreaRunsOnWarpbuildBlacksmithNamespaceUbicloud

Median across providers: 4 → 8 vCPU: 29% faster, +40% cost (6 providers)

Beyond builds

The same runners, one question each.

  • CPU Fastest single-thread on x64: Namespace 4x8, 4,700 PassMark.
  • Cache 4 GB restore on x64, any size: RunsOn 10s (12 vCPU), then Blacksmith 13s (8 vCPU); GitHub's own service 28s (8 vCPU).4 GB restore on x64 at 2 vCPU: Blacksmith 14s, then Avrea 15s; GitHub's own service 50s.4 GB restore on x64 at 4 vCPU: Blacksmith 14s, then Avrea 14s; GitHub's own service 34s.4 GB restore on x64 at 8 vCPU: RunsOn 13s, then Blacksmith 13s; GitHub's own service 28s.
  • Disk Small random reads (QD1) on x64: Blacksmith 2 vCPU leads with 80,883 per second, 46× the slowest disk (AWS CodeBuild, 1,766). That is a disk's limit, and a build often never reaches it.
  • EC2 storage NVMe, gp3 and tmpfs on the same instances, and what each does to the builds above.
  • tmpfs The workspace in RAM instead of on a disk: what it does to the builds above, what it costs and where it does nothing.
  • CPU Fastest single-thread on arm64: Avrea 2 vCPU, 5,028 PassMark.
  • Cache 4 GB restore on arm64, any size: Avrea 11s (8 vCPU), then RunsOn 13s (8 vCPU); GitHub's own service 29s (4 vCPU).4 GB restore on arm64 at 2 vCPU: Avrea 12s, then Warpbuild 28s; GitHub's own service 52s.4 GB restore on arm64 at 4 vCPU: Avrea 12s, then RunsOn 18s; GitHub's own service 29s.4 GB restore on arm64 at 8 vCPU: Avrea 11s, then RunsOn 13s.
  • Disk Small random reads (QD1) on arm64: Warpbuild xfast (Apple M4 Pro) 6 vCPU leads with 26,620 per second, 3.2× the slowest disk (GitHub, 8,202). That is a disk's limit, and a build often never reaches it.
  • EC2 storage NVMe, gp3 and tmpfs on the same instances, and what each does to the builds above.
  • tmpfs The workspace in RAM instead of on a disk: what it does to the builds above, what it costs and where it does nothing.

Providers

Price per minute of a 4 vCPU runner on x64 (its shape beside each price); RunsOn at EC2 spot and on-demand.Price per minute of a 4 vCPU runner on arm64 (its shape beside each price); RunsOn at EC2 spot and on-demand.

Head-to-head

Two runners of the same size, build for build. Cost per build decides: a faster runner that costs twice as much per build is not the better runner.

Cost per build is variable cost only; licenses and plan fees are flat, on each provider's page. Estimate your bill