Reviewed against primary sources May 2026
About cicdpipelinecost.com
An independent reference for what a CI/CD pipeline actually costs in May 2026: an interactive cost calculator, a neutral six-provider external-runner shootout, and a seven-step engineering playbook for cutting CI spend 40 to 90%. No vendor relationships, no quote forms.
Why this site exists
CI/CD vendor pricing pages are scattered, use different units (minutes, credits, agents, build hours, parallel jobs), and sit inside marketing funnels that lead with sales contact forms before per-minute numbers. The GitHub Actions macOS 10x multiplier alone takes three different vendor docs pages to translate into a per-minute rate. The CircleCI credit model converts to a per-minute number only after looking up the resource-class table. The Buildkite orchestration-only model needs careful read-through to separate orchestration fees from compute. The external-runner market (Depot, Namespace, BuildJet, Blacksmith, Ubicloud, RunsOn) has six pitches and zero neutral comparisons.
This site reduces those pages to comparable per-minute figures with the methodology shown, a calculator that models a real workload, and a neutral runner shootout that compares external providers on the same axes. The audit goal is reproducibility: every number on the site should be re-derivable from the vendor's own public pricing page or a cited engineering source.
CI is the canonical unit-cost problem in engineering. It is invisible, it scales with team size, and it is optimisable. The seven-step optimisation playbook on the homepage covers the full range from a 30-minute paths filter to a one-week external-runner migration. Most teams recover 40 to 90% of wasted spend within two weeks.
Who builds this
cicdpipelinecost.com is built and maintained by Oliver Wakefield-Smith at Digital Signet, an independent reference-content studio. The site is part of a four-publication tech-editorial series covering the cost layers of modern engineering organisations: featurebloat.com (product layer), codesmellcost.com (engineering layer), contextcost.com (attention layer), and cicdpipelinecost.com (build-pipeline layer).
For pure pricing-reference coverage of CI/CD vendors, see the sister site cicdcost.com, which carries dedicated per-vendor pricing pages for GitHub Actions, CircleCI, GitLab CI, Buildkite, AWS CodeBuild, Bitbucket Pipelines, Drone CI, and Jenkins. For workload-modelled scenario analysis, see cicdcalculator.com. The three sites cover distinct intent slots: pricing reference (cicdcost), scenario calculator (cicdcalculator), optimisation playbook and runner shootout (cicdpipelinecost).
Related Digital Signet cost-reference properties: platformengineeringcost.com, monitoringcost.com, egresscost.com, apigatewaycost.com.
Editorial position
This is a reference site, not a reseller, not a managed-services lead-generation property, and not a vendor advocacy outlet. The runner shootout is constructed neutrally: every provider scored on the same axes, no preferential ordering, and no provider gets to write its own row. Where a non-affiliate vendor (Ubicloud) beats an affiliate vendor on a given axis, we say so.
Where pricing context is contested between sources, both ends of the range are shown with the assumption stated. Where vendor pricing pages drift between verification cycles, the new number is published with a single-source date stamp, not silently swapped in. Where a published case study is available (Dropbox / Bazel, Spotify, AirBnB, Shopify) we link to the original engineering blog post rather than paraphrase.
Cited primary sources include: vendor pricing pages (GitHub Actions billing docs, GitLab pricing, CircleCI pricing, Atlassian Bitbucket Pipelines, AWS CodeBuild and CodePipeline pricing, Azure Pipelines, Google Cloud Build, JetBrains TeamCity, Buildkite, GoCD), external-runner-provider pricing pages (Depot, Namespace, BuildJet, Blacksmith, Ubicloud, RunsOn), the DORA State of DevOps report, Forrester / Gartner CI/CD market data, the CNCF Cloud Native developer survey, and engineering-blog post-mortems from samexpert.com, northflank.com, and blacksmith.sh.
What this site covers
Editorial principles
Every per-minute rate and free-tier number on this site traces back to a vendor's own public pricing page (GitHub, GitLab, CircleCI, Atlassian, AWS, Microsoft Azure, Google Cloud, Buildkite, JetBrains TeamCity, Depot, Namespace, BuildJet, Blacksmith, Ubicloud, RunsOn). Where a vendor uses a credit or minute-multiplier model the conversion is shown alongside the published rate.
The Depot vs Namespace vs Blacksmith table is published side by side with the same methodology applied to every provider: per-minute rate, ARM rate, macOS support, cold start, cache mechanism, setup time, minimum commit. No vendor was given preferential ordering. We are not a reseller for any runner provider.
Some external-runner links carry affiliate parameters where the provider operates an affiliate programme; the disclosure banner is shown on every page. Affiliate participation does not change ranking, pricing accuracy, or what we recommend. Where a non-affiliate vendor offers a better deal than an affiliate vendor we say so.
Pricing is re-verified against each vendor's own pricing page on the first business week of each month. The last verified label currently reads May 2026.
The verification date is held in one constant (LAST_VERIFIED_DATE) imported by every page. Footer text, schema dateModified, and visible headings all read from that single source so cosmetic refreshes are not possible.
Self-hosted runner break-even math uses the published GitHub-hosted per-minute rate, a published AWS EC2 on-demand instance rate, and a maintenance-time estimate from published industry sources (DORA State of DevOps reports, public engineering-blog post-mortems). Spot, savings plan, and reserved-capacity discounts are explicitly out of scope to keep the comparison apples-to-apples.
Methodology in brief
Per-minute rates come from vendor public pricing pages. Credit models (CircleCI) and minute-multipliers (GitHub Actions Windows 2x, macOS 10x) are translated to per-minute rates with the conversion shown. Self-hosted break-even math uses the published GitHub-hosted rate, an AWS EC2 on-demand instance rate (t3.medium reference), and a maintenance-hours estimate published in DORA reports and engineering-blog post-mortems. Both ends of the maintenance-hours range are shown.
The runner shootout scores Depot, Namespace, BuildJet, Blacksmith, Ubicloud, and RunsOn on identical axes: Linux x86 per-minute rate, Linux ARM per-minute rate, macOS support, cold start, cache mechanism, setup time, minimum commit. The shootout table indicates the best and worst values in each column.
For full source provenance, calculation framework, in-scope / out-of-scope coverage, and the corrections process, see the methodology page.
Contact and corrections
Spotted a stale rate, a missing tier, a runner provider we have not covered yet, or a vendor change we have not caught? Email [email protected] with the page URL and the source you would like cited. Substantive corrections are typically actioned within five business days.
Disclosures
- ●Some external-runner links carry affiliate parameters (Depot, Namespace, BuildJet, Blacksmith, RunsOn). Disclosure banner is shown on every page.
- ●Not affiliated with GitHub, GitLab, CircleCI, Buildkite, Atlassian, Amazon Web Services, Microsoft, Google Cloud, JetBrains, or any other CI/CD vendor.
- ●No email-gated downloads, quote forms, or sales redirects on any page.
- ●Calculator outputs are estimates; production pricing depends on enterprise agreements, regional surcharges, and reserved-capacity commitments not covered here.