How we make this list: we compare pricing pages, product documentation and public user reviews. Ratings come from G2, Capterra or Trustpilot, with the date we checked them.
Key takeaways
- GitHub Actions is the default for teams on GitHub: free for public repositories, 2,000 free minutes a month on private ones, a marketplace of reusable actions including every major cloud and Terraform workflow, and self hosted runners when builds must stay inside your network.
- GitLab CI/CD is the all in one: code, pipelines, registry and security scanning in one platform, free to start, Premium at $29 per user per month, and the strongest single vendor answer for regulated teams.
- Jenkins and Argo CD are the free, self hosted poles: Jenkins for total build control with a vast plugin ecosystem, Argo CD for GitOps delivery to Kubernetes only, and both bill you in operations time instead of licences.
- CircleCI, Harness, AWS CodePipeline and Azure Pipelines cover the specialist cases: raw build speed, enterprise release governance, and pipelines native to one cloud.
Quick comparison
Is your product missing or out of date?
We review this list regularly. Vendors can request a listing, send updated pricing or ask about featured placement.
Starting priceFree
Free planYes
GitHub Actions wins by absence of decision. The code is already there, the team already has accounts, and the first pipeline is a file in the repo. For cloud infrastructure work its marketplace matters most: plan and apply Terraform, deploy to every major cloud, scan configurations, all as maintained actions rather than scripts you own.
The trade is metering and sprawl. Minutes are cheap until an infrastructure monorepo runs a forty minute plan on every pull request, and the 2026 platform fee made private repo pipelines worth auditing. Mature teams respond with reusable workflows, path filters and self hosted runners for the heavy jobs.
What we like- Zero adoption friction for GitHub teams: workflows live in the repository as YAML, triggered by the events the team already uses, free for public repositories, with 2,000 Linux minutes a month on the free plan for private ones, the largest action marketplace in the category, including official cloud deploy and Terraform actions, self hosted runners keep builds and cloud credentials inside your own network
Watch out for- Billing needs watching: private repositories added a per minute platform fee on top of compute in March 2026 per current reporting, so heavy pipelines are a metered cost, workflow YAML sprawls as estates grow, and reusable workflow discipline is on you, GitHub coupling is total: leaving GitHub means rewriting every pipeline
Starting priceFree
Free planYes
GitLab is the considered choice rather than the default one: an organisation picks it because a single system of record for code, builds, scans and deployments simplifies audits and access control. Its infrastructure support is genuine, managed Terraform state and a Kubernetes agent built for GitOps style delivery.
The trade is commitment and cost per seat. A 200 developer organisation pays real money at Premium, and pays it to one vendor for most of its delivery chain. Teams that want that consolidation call it efficiency; teams that prefer best of breed tools call it lock in. Both are describing the same contract.
What we like- One platform for repositories, pipelines, container registry, security scanning and compliance frameworks, which regulated industries value more than any single feature, free tier includes 400 compute minutes a month on shared runners
- Premium at $29 per user per month brings 10,000 minutes and the governance features, self managed deployment keeps everything, runners and credentials, inside your own infrastructure, strong native support for Terraform state and Kubernetes agents
Watch out for- The all in one is also an all in migration: adopting GitLab CI usually means adopting GitLab, per user pricing at Premium and Ultimate tiers, $99, is the category's steepest for large teams, the platform's breadth means a deeper administration surface than a pipeline only tool
#3CircleCI
Best for: Fast managed builds4.4/5 on
G2 (510 reviews, checked Oct 2026)
Visit circleci.com
Starting priceFree
Free planYes
CircleCI is bought by teams whose pipelines have become the bottleneck, monorepos, heavy integration tests, infrastructure plans across many accounts, where shaving ten minutes per run returns engineering hours weekly. Its caching and parallelism are the most polished in the managed field.
The trade is the meter. Credits make costs a function of your own configuration choices, which rewards teams that measure and punishes teams that set and forget. For an infrastructure pipeline that runs rarely, CircleCI is often over specified; for one that runs a hundred times a day, it pays for itself in recovered time.
What we like- Build speed is the product: strong caching, parallelism and large resource classes that cut pipeline times for big test suites and infrastructure validation, a generous free plan with 30,000 credits a month, and open source projects can now draw up to 400,000 credits monthly in current reporting, Performance plans start at $15 a month plus usage, self hosted runners available when jobs must run inside your network
Watch out for- Credit based pricing is the category's least predictable: resource class choices change burn rates, and bills surprise teams that do not instrument them, a pipeline only tool: repositories, registry and security live elsewhere, by design, configuration depth is real
- Extracting its speed takes pipeline engineering, not defaults
#4Jenkins
Best for: Self hosted build control4.4/5 on
G2 (570 reviews, checked Oct 2026)
Visit jenkins.io
Starting priceFree
Free planYes
Jenkins remains the answer where control is non negotiable: unusual hardware, strict network rules, a decade of build scripts encoding how the company actually ships. It will run a Terraform plan, a mainframe compile and a Kubernetes deploy in one pipeline, because a plugin exists for each.
The trade is honesty about staffing. Jenkins is free the way a puppy is free. Organisations with a platform team run it happily for years; a five person startup adopting it in 2026 is volunteering for a second job.
What we like- Free, open source and entirely yours: the most deployed CI server in the world, with a plugin ecosystem covering every cloud, tool and odd build environment ever shipped, runs anywhere, including air gapped and regulated networks no SaaS tool may enter, no per minute or per user vendor bill, ever
- The costs are your infrastructure and your time
Watch out for- You are the vendor: upgrades, plugin compatibility, controller security and scaling agents are your team's work, the plugin model that gives Jenkins its reach is also its maintenance burden and its historical security surface, cloud native alternatives win most new projects, per current comparisons, because managed setup beats control for standard work
Starting priceFree
Free planYes
Argo CD is listed because for Kubernetes estates the CD half of CI/CD deserves its own specialist. The model, desired state in Git, an agent that enforces it, is the cleanest answer yet devised to who changed production and how do we roll back: the commit history answers both.
The trade is scope and dogma. Argo CD manages what Kubernetes manages; the databases, queues and cloud accounts around the cluster still need Terraform and a pipeline, and the full stack becomes Argo plus a CI tool plus discipline.
What we like- The GitOps standard for Kubernetes delivery: Argo CD continuously reconciles the cluster's actual state with the manifests in Git, so the repository is the deployment record, drift is visible and sync is auditable, free and open source under Apache 2.0, run in cluster, declarative everything suits infrastructure teams: environments are branches and folders, promotion is a pull request
Watch out for- It deploys
- It does not build or test, so a CI tool from this list still feeds it, Kubernetes only: no answer for VMs, serverless estates or the cloud resources outside the cluster, GitOps discipline is all or nothing, and teams that hand patch clusters will fight its reconciliation until they stop
#6Harness
Best for: Enterprise release governance4.6/5 on
G2 (336 reviews, checked Oct 2026)
Visit harness.io
Starting priceSee website
Free planYes
Harness sells the part of delivery that frightens large organisations: the promotion of a change through environments with evidence, approvals and an automatic retreat when production disagrees. Its continuous verification, watching metrics and logs after a deploy and rolling back on anomaly, is the feature set its rivals cite most.
The trade is weight. A twenty person company does not have the release governance problem Harness solves, and will pay enterprise prices to acquire process it does not need. A two thousand person company with auditors has the opposite problem, and for it the quote based pricing is simply how such tools are bought.
What we like- Built for the enterprise release problem: deployment pipelines with approvals, policy checks, automated verification against monitoring data and rollback when the metrics say the release is bad, a free tier covers small teams, with paid tiers sold on quote, strong fit for multi cloud estates where releases cross AWS, Azure and on premise in one flow, governance and audit features are first class, not bolted on
Watch out for- Enterprise pricing opacity: the real tiers require a sales conversation, the platform has grown by acquisition into many modules, and buyers must check which their quote includes, verification and rollback magic depends on integrating your observability stack properly, which is a project, not a toggle
Starting price$1 per pipeline/month
Free planNo
CodePipeline is the choice when the estate is AWS and the governance answer to everything is IAM. Security teams like it for a structural reason: the pipeline's powers are policies they already review, in the console they already audit, with no third party holding cloud credentials.
The trade is experience and reach. Engineers accustomed to Actions find CodePipeline's model rigid, and any workload outside AWS needs a second system anyway. It is best understood as the delivery layer of the AWS platform rather than a competitor to the general tools, and judged on those terms it is hard to beat on price or compliance.
What we like- The native answer inside AWS: pipelines wired directly into IAM, CodeBuild, CloudFormation, ECS and Lambda, with permissions expressed in the same policy language as everything else you run, pricing is close to negligible: $1 per active pipeline per month on V1 after the free first month, or per action execution minutes on V2 in current pricing, no separate vendor, contract or credential store: the pipeline is an AWS resource, audited by CloudTrail like the rest
Watch out for- AWS only, and contentedly so: multi cloud estates get a partial tool, the developer experience and pipeline modelling trail GitHub Actions and GitLab for complex builds, most teams use it as an orchestrator over CodeBuild or external CI rather than loving it directly
Starting priceFree
Free planYes
Azure Pipelines closes the list as the incumbent for Microsoft estates: organisations standardised on Azure and .NET get identity, boards, repos and pipelines in one tenant, with Windows build agents that are genuinely first class.
The trade is gravitational. Microsoft itself positions GitHub Actions as the future for much CI, and new Azure teams increasingly start there. Pipelines remains the right answer where Azure DevOps is the system of record, or Windows heavy builds dominate.
What we like- The Microsoft stack's native pipeline: deep integration with Azure services, Boards and Repos inside Azure DevOps, plus full support for GitHub hosted code, a free tier with one hosted parallel job and 1,800 minutes a month, additional hosted jobs at $40 a month each in current pricing, YAML or classic pipelines, with self hosted agents for builds that must run on your machines, strong Windows and .NET build support, the historical Jenkins ground
Watch out for- The Azure DevOps surface is several products wearing one name, and new users meet that complexity early, outside the Microsoft ecosystem it has no pull: GitHub Actions eats its general purpose lunch, including for many Azure deployments, per job pricing for parallelism is simple but blunt compared with per minute rivals
How we chose
We required real cloud infrastructure relevance: providers, Terraform and Kubernetes workflows, self hosted runner options and identity integration, then spread the list across hosted CI, all in one platforms, open source control and cloud native delivery. Prices are quoted in each vendor's native unit from current 2026 pricing pages and dated comparisons, flagged where models changed recently, as GitHub's private repo fees did. CD only specialists are labelled as such. No hands-on testing is claimed.
Pairing CI with delivery for infrastructure
Most cloud estates end up with two layers from this list, and choosing the pair deliberately beats growing it accidentally. The CI layer, Actions, GitLab, CircleCI or Jenkins, builds artefacts, runs tests, and for infrastructure produces plans: a Terraform plan on every pull request, reviewed like code, applied on merge with the same credentials and state locking every time.
The delivery layer decides how change reaches the cloud. For Kubernetes workloads, Argo CD holding desired state in Git is the pattern of the moment, fed by CI that publishes images and updates manifests. For cloud resources themselves, the pipeline is the delivery layer, which is why IAM design matters more than tool choice: the role your apply step assumes should be able to do the change and little else, scoped per environment.
Three failure stories repeat across postmortems. Pipelines holding standing admin credentials, so a compromised workflow owns the cloud. Plans reviewed in a hurry because the tool made apply one click and plan a scroll past. And no rehearsed rollback for infrastructure changes, which unlike app deploys cannot always be undone by shipping the previous version. Pick tools that make the safe path the easy path, short lived credentials, plan artefacts attached to the pull request, state backed up and versioned.
Frequently asked questions
What is the difference between CI and CD for infrastructure?
Continuous integration builds and tests change, including validating infrastructure code, linting Terraform, scanning configurations, producing a plan. Continuous delivery promotes approved change into environments, and continuous deployment does it automatically. For infrastructure the approval gate survives longer than for apps, because an apply can alter or destroy resources, so most teams run delivery with a human merge as the gate.
Which tool is best for Terraform pipelines?
The one your code already lives beside, for most teams: GitHub Actions and GitLab CI both have mature Terraform workflows, plan on pull request, apply on merge, managed state in GitLab's case. Jenkins serves estates with existing investment. The tool matters less than the pattern: remote state with locking, short lived cloud credentials via OIDC, and a plan artefact a reviewer reads.
Should we use Argo CD and GitHub Actions together?
Yes, that is the common modern stack for Kubernetes: Actions builds and tests, publishes the image, and updates the manifests in Git; Argo CD reconciles the clusters to those manifests and reports drift. Each does the half it is best at. The discipline cost is keeping Git as the only path to the cluster, which is the point of the model.
How do the pricing models compare in real money?
A ten developer team in current comparisons pays roughly nothing on GitHub Actions free minutes if builds are light, about $290 a month on GitLab Premium, $15 a month plus credit usage on CircleCI Performance, and infrastructure plus engineering time for Jenkins or Argo CD. The honest variable is build minutes: measure your current usage before comparing any units.
Are free and open source tools production grade?
Jenkins and Argo CD run some of the largest estates in the world; their cost is operations, not capability. The real question is whether your team wants to own upgrades, security patches and scaling for the delivery system itself. Managed tools convert that work into a subscription.
Sources we checked (0)
Is your product missing or out of date?
We review this list regularly. Vendors can request a listing, send updated pricing or ask about featured placement.
Shihab Mia · Founder, Rankite
Shihab has 5+ years in B2B and SaaS SEO. He writes Rankite's software buying guides from pricing pages, documentation and public user reviews. More about Shihab.
GitHub Actions
GitLab CI/CD
Argo CD
AWS CodePipeline
Azure Pipelines