Vitko Runners vs GitHub-hosted (ubuntu-latest): speed and cost on 12 real CI workloads

Speed

× faster than GitHub-hosted (ubuntu-latest) · higher is better

Vitko Runners (warm starts + test splitting)

2.3×

Vitko Runners

1.7×

GitHub-hosted (ubuntu-latest)

1.0×

USD per $1.00 spent on GitHub-hosted (ubuntu-latest), at 100,000 jobs a month · lower is better

Vitko Runners (warm starts + test splitting)

$0.16

Vitko Runners

$0.16

GitHub-hosted (ubuntu-latest)

$1.00

Vitko Runners had the lowest median time of the two on 12 of 12 workloads, GitHub-hosted (ubuntu-latest) on 0.

Workload	Vitko Runners (warm starts + test splitting)	Vitko Runners	GitHub-hosted runners (ubuntu-latest, 2 vCPU)	Lowest median
Burst: 20 jobs at once	131 s	160 s	259 s	Vitko Runners (warm starts + test splitting)
cli	140 s	272 s	368 s	Vitko Runners (warm starts + test splitting)
codex-lint	405 s	404 s	884 s	Vitko Runners
container	178 s	179 s	239 s	Vitko Runners (warm starts + test splitting)
fd	45 s	49 s	92 s	Vitko Runners (warm starts + test splitting)
flask	13 s	16 s	23 s	Vitko Runners (warm starts + test splitting)
guava	651 s	655 s	1167 s	Vitko Runners (warm starts + test splitting)
monorepo	172 s	180 s	384 s	Vitko Runners (warm starts + test splitting)
pydantic	69 s	105 s	179 s	Vitko Runners (warm starts + test splitting)
ts	243 s	372 s	784 s	Vitko Runners (warm starts + test splitting)
vite	199 s	350 s	603 s	Vitko Runners (warm starts + test splitting)
vue	32 s	63 s	121 s	Vitko Runners (warm starts + test splitting)

Mean cost per job in USD, from the benchmark’s sensitivity results

Volume	Vitko Runners (warm starts + test splitting)	Vitko Runners	GitHub-hosted runners (ubuntu-latest, 2 vCPU)
100,000 jobs a month	$0.158	$0.157	$1.000
10,000 jobs a month	$0.158	$0.157	$1.000
1,000 jobs a month	$0.158	$0.157	$1.000

Counted, never replaced

Vitko Runners (warm starts + test splitting): 3 of 360 measured runs (0.8%)
Vitko Runners: 0 of 360 measured runs (0.0%)
GitHub-hosted runners (ubuntu-latest, 2 vCPU): 3 of 360 measured runs (0.8%)

Each configuration page lists where its failures happened. All losses.

One line per configuration pairing, generated from the benchmark data

Vitko Runners (warm starts + test splitting) ran jobs 2.3× faster than GitHub-hosted runners (ubuntu-latest, 2 vCPU) (95% interval 2.2× to 2.4×); Vitko Runners (warm starts + test splitting) cost 84% less per job (at 100,000 jobs a month).
Vitko Runners ran jobs 1.7× faster than GitHub-hosted runners (ubuntu-latest, 2 vCPU) (95% interval 1.7× to 1.8×); Vitko Runners cost 84% less per job (at 100,000 jobs a month).