Cut your CI bill without cutting corners

Tactical, workflow-level fixes for the most common GitHub Actions cost problems. Each guide covers symptoms, metrics to watch, and exact YAML changes.

CI runs too often

Reduce run frequency

Highest-ROI category. Fixes here immediately reduce billable minutes.

Why your CI runs on every push concurrency groups, auto-cancel Duplicate runs from misconfigured triggers push + pull_request overlap Docs changes trigger full CI path filters Monorepo: every change runs everything path-based filtering Scheduled jobs that don't earn their keep cron audit, event-driven Canceled runs are wasting minutes concurrency, debounce Control ChatOps/bot-triggered CI actor filtering

Too much work per run

Trim work per execution

Workflow architecture changes that reduce what each CI run actually does.

Matrix explosion reduce matrix on PRs Too many small jobs, too much overhead job consolidation Build once, use everywhere artifacts, build-first Slow failures: expensive steps before cheap ones needs keyword, fast-fail E2E tests running too often risk gating, merge queue Separate release/deploy from CI gate to main/tags Optimize lint/typecheck ROI incremental, changed-files Reduce setup & install overhead caching, prebuilt images Optimize security scans diff-aware, scan on main Speed up CI pipelines slow jobs, parallelism Optimize CI tests on PRs test splitting, path filters Find and fix CI cost hotspots workflow cost concentration

Recomputing the same work

Stop redoing cached work

CI redoes work that could be reused. Often invisible until measured.

Dependency cache not working cache keys, restore strategy Docker builds never cached BuildKit, layer ordering Hidden cost of huge build contexts .dockerignore, shallow clone

Reruns & flakiness

Kill the retry tax

CI spends money on non-determinism instead of testing code.

Flaky tests are costing you real money quarantine, partial reruns Over-parallelized test suites shard sizing, duration split Integration tests rebuilding the world schema load, tx rollback Reduce CI timeouts timeout-minutes Reduce CI failures infra vs test failures Stabilize CI runtime variance p90/p50, cache hit rate

Artifacts & storage

Cut storage bloat

CI generates and stores more data than needed.

Artifacts uploaded on every run conditional upload, retention Caching too much (and paying for it) audit, split, prune

Runner & infrastructure mismatch

Right-size your runners

Wrong runner size or type inflates cost per minute.

Underpowered runners right-size CPU cores Overpowered runners for lightweight jobs match runner to workload Always-on self-hosted runners autoscale, scale-to-zero Reduce macOS/Windows CI spend OS matrix gating, 10x rate Queue time exceeds job runtime concurrency, job consolidation

Reference

Pricing & rate cards

GitHub Actions pricing & per-minute rates rate cards, multipliers How GitHub Actions billing works metering, included minutes, overages

Comparisons

Platform comparisons

GitHub Actions vs GitLab CI pricing, config, runners GitHub Actions vs CircleCI credits, orbs, executors GitHub Actions vs Jenkins YAML vs Groovy, hosted vs self-managed GitHub Actions vs Azure DevOps per-minute vs parallel jobs, templates GitHub Actions vs Google Cloud Build GCP integration, machine types, caching

$ costops analyze --workflows

Stop guessing which workflows waste money

CostOps shows you per-workflow cost, run frequency, and failure rates, so you know exactly where to apply these fixes.

Get Started with GitHub

Free for 1 repo. No credit card. No code access.

Built by engineers who've managed CI spend at scale.