Backstage vs Port vs Cortex: Which Internal Developer Portal in 2026?
Backstage, Port, and Cortex compared for internal developer portals in 2026 — self-hosting effort, catalog setup time, scorecards, plugin ecosystem, and which one actually gets adopted by engineers.
Every platform team eventually asks the same question: build on Backstage, or buy Port/Cortex. Here is an honest comparison based on what it actually takes to get an IDP adopted, not just installed.
Quick Comparison
| Backstage | Port | Cortex | |
|---|---|---|---|
| Model | Open source, self-hosted | SaaS (+ self-hosted option) | SaaS |
| Setup time | Weeks (plugin config, catalog YAML) | Hours to days | Days |
| Software catalog | YAML-based, manual entity registration | UI + API + auto-discovery | API + integrations, auto-discovery |
| Scorecards | Requires plugin (Tech Insights) | Built-in, first-class | Built-in, first-class |
| Plugin ecosystem | Huge (500+ open source plugins) | Growing marketplace | Smaller, curated |
| Cost | Free (self-host infra cost) | Paid, free tier for small teams | Paid |
| Ownership | Full control, full maintenance burden | Vendor manages upgrades | Vendor manages upgrades |
Backstage
Backstage is Spotify's open-source developer portal framework — a React app you extend with plugins, backed by a software catalog defined in YAML.
# catalog-info.yaml — lives in each service repo
apiVersion: backstage.io/v1alpha1
kind: Component
metadata:
name: payments-api
annotations:
github.com/project-slug: myorg/payments-api
pagerduty.com/service-id: PXXXXX
spec:
type: service
lifecycle: production
owner: team-payments
system: checkoutBackstage strengths:
- Free and fully self-hosted — no per-seat cost at scale
- Enormous plugin ecosystem: CI/CD status, cost insights, TechDocs, Kubernetes, SonarQube
- Full control over data model, auth, and UI customization
- Backed by CNCF, large community, most enterprises already have someone running it
Backstage weaknesses:
- Real setup cost is weeks, not hours — plugin config, catalog population, and auth wiring all need engineering time
- Someone owns the upgrade and maintenance burden forever (this is the real hidden cost)
- Catalog entities require manual YAML per service unless you build auto-discovery yourself
- UI feels dated compared to Port/Cortex out of the box
When to use Backstage: You have a platform team that can dedicate an engineer or two to build and maintain it, and you want zero vendor lock-in.
Port
Port is a SaaS internal developer portal focused on getting a working catalog and scorecards live fast, with a self-hosted option for regulated environments.
# port-app-config.yml — blueprint definition
identifier: service
title: Service
properties:
language:
type: string
owner:
type: string
productionReadiness:
type: number
relations:
repository:
target: githubRepositoryPort strengths:
- Fastest time-to-value — auto-discovery from GitHub/GitLab gets a populated catalog in hours
- Scorecards and self-service actions are first-class, not bolted on
- No-code UI for defining blueprints — non-platform engineers can extend it
- Good workflow automation (self-service actions trigger Terraform, GitHub Actions, etc.)
Port weaknesses:
- Paid beyond a small free tier — cost scales with team size
- Less deep customization than Backstage's plugin model for unusual use cases
- Smaller plugin/integration list than Backstage's community ecosystem
When to use Port: You want an IDP live and adopted within a sprint, not a quarter, and you are fine paying for that speed.
Cortex
Cortex focuses heavily on scorecards and engineering maturity tracking — think "service quality dashboard" more than "developer portal."
# cortex.yaml — service definition
title: Payments API
x-cortex-tag: payments-api
x-cortex-git:
github:
repository: myorg/payments-api
x-cortex-service-groups:
- checkoutCortex strengths:
- Scorecards are the best in class — clear engineering standards tracking with automated grading
- Strong integrations for on-call, incident management, and CI/CD signal aggregation
- Good for driving org-wide initiatives ("every service must have a runbook by Q3")
Cortex weaknesses:
- Less of a general-purpose portal than Backstage or Port — thinner on self-service actions
- SaaS-only pricing scales similarly to Port
- Catalog customization is more rigid than Backstage's fully open data model
When to use Cortex: Your primary pain is inconsistent service quality across teams and you need scorecards to drive accountability, not a broad developer portal.
The Honest Verdict
Have a dedicated platform team and want zero vendor lock-in: Backstage. Budget weeks for setup and an ongoing maintenance owner — that cost is real and often underestimated.
Need adoption fast and can pay for it: Port. It gets a populated catalog and self-service actions live faster than anything else here.
Your specific pain is engineering standards and scorecards: Cortex. It does that one thing better than the generalists.
Most mid-size teams underestimate Backstage's maintenance cost and overestimate Port/Cortex's pricing at their actual team size — run the math on both before committing.
More platform engineering? Read our Platform engineering: the future of DevOps and How to set up Backstage developer portal.
Today I Fixed
Short real fixes from production — posted daily
Stay ahead of the curve
Get the latest DevOps, Kubernetes, AWS, and AI/ML guides delivered straight to your inbox. No spam — just practical engineering content.
Related Articles
Coolify vs CapRover vs Dokploy: Which Self-Hosted PaaS in 2026?
Coolify, CapRover, and Dokploy compared for self-hosted deployment platforms in 2026 — setup time, Docker Compose support, database management, one-click apps, and which to run on your own VPS.
Gitea vs Self-Hosted GitHub Enterprise vs GitLab: Which in 2026?
Gitea, GitHub Enterprise Server, and self-hosted GitLab compared for 2026 — resource footprint, CI/CD built-in vs bolt-on, feature completeness, and which to pick for running your own Git hosting.
A2A Protocol 1.0: Why Agent-to-Agent Communication Is Becoming a Standard
A practical guide to A2A Protocol 1.0, agent discovery, task collaboration, and why A2A complements MCP instead of replacing it.