Platform engineering has a tooling problem. The category has grown fast since 2022, and the number of products competing for your attention has grown faster. If you search for "platform engineering tools" today, you'll find developer portals, platform orchestrators, service catalogs, maturity scorecards, and self-service frameworks — all described using overlapping language, all claiming to solve the internal developer platform problem.

The tools in this space fall into three categories that matter more than individual feature lists: developer portals that provide visibility into your existing stack, platform orchestrators that standardize how infrastructure gets provisioned, and developer platforms that replace parts of the stack entirely. Picking from the wrong category is a more expensive mistake than picking the wrong tool within the right one.

This comparison covers the six tools that come up most often when teams evaluate their options: Backstage, Port, Cortex, OpsLevel, Humanitec, and Encore Cloud.

Quick Comparison

Tool	Category	Hosting	Pricing Model	Service Catalog	Infra Provisioning	CI/CD	Observability
Backstage	Developer portal	Self-hosted	Free (OSS) + team cost	Yes	No	No (plugin)	No (plugin)
Port	Developer portal	SaaS	Free tier, then per user	Yes	No (triggers external)	No (triggers external)	No
Cortex	Standards & maturity	SaaS	Enterprise pricing	Yes	No	No	No
OpsLevel	Service ownership	SaaS	Per user	Yes	No	No (checks only)	No
Humanitec	Platform orchestrator	SaaS / self-hosted	Custom pricing	Partial	Yes (orchestration)	Partial	No
Encore Cloud	Developer platform	SaaS	Free tier, then per member	Yes (auto-generated)	Yes (automatic)	Yes (built-in)	Yes (built-in)

Backstage

Backstage is Spotify's open-source developer portal, open-sourced in March 2020 and later accepted into the CNCF Sandbox. It gives you a framework for building an internal developer portal with a plugin architecture. The core ships with a service catalog, documentation system (TechDocs), and software templates. Everything else — Kubernetes status, CI/CD integration, cloud cost dashboards, API documentation — comes through plugins.

Pricing: The software is free. The cost is the engineering time to deploy, customize, and maintain it. Most organizations that run Backstage in production dedicate one to three engineers to it. Roadie offers managed Backstage at $22/developer/month if you want to skip the operational overhead.

Strengths: The plugin ecosystem is the largest of any developer portal. You can build almost anything into Backstage if you have React developers and time. Spotify runs thousands of services through it, which is real validation for the architecture. Because it's open source, there's no vendor dependency on the portal layer itself.

Limitations: Backstage is a portal framework, not a finished product. A production deployment takes weeks of configuration, and ongoing maintenance is substantial — Backstage releases frequently, and plugin compatibility breaks on upgrades. The bigger constraint is architectural: Backstage displays information about your infrastructure but doesn't manage it. You still need Terraform, Kubernetes, ArgoCD, Datadog, and the rest of your stack. Backstage adds a UI on top of that complexity rather than reducing it.

Best for: Organizations with 200+ engineers, a dedicated platform team, and an existing infrastructure toolchain that needs a unified frontend.

Port

Port takes Backstage's concept and delivers it as a managed SaaS product. Instead of writing plugins, you configure your data model through a visual interface: define entities (services, environments, cloud resources, packages), connect integrations, and build self-service actions that trigger workflows in your existing CI/CD systems.

Pricing: Free for up to 15 users. Paid plans start around $30/seat/month and scale based on features and user count. Enterprise pricing is custom.

Strengths: Time-to-value is significantly faster than Backstage. You configure rather than code, so a single platform engineer can get a working portal running in days rather than weeks. Scorecards let you track engineering standards across services (does it have a README, is it using the approved base image, when was the last deployment). Self-service actions are the standout feature — you can build workflows like "create a new microservice" or "spin up a dev environment" that trigger GitHub Actions, Jenkins, or whatever pipeline you're running.

Limitations: Port sits on top of your existing toolchain, the same as Backstage. The self-service actions are powerful, but they're wrappers around your CI/CD — Port doesn't provision infrastructure or run deployments itself. If you don't already have GitHub Actions or ArgoCD configured to handle those workflows, Port can't create them for you. SaaS pricing also adds up at scale; a 50-person engineering team is paying $18,000+/year before reaching the enterprise tier.

Best for: Teams that want Backstage's portal capabilities without running Backstage, and who already have a working infrastructure and CI/CD stack underneath.

Cortex

Cortex focuses narrower than Backstage or Port. Its core value proposition is engineering standards enforcement: you define scorecards with criteria (service has runbooks, uses approved languages, has SLOs configured, passed its last security scan), and Cortex shows you which services comply and which don't.

Pricing: Enterprise-oriented, pricing is not public. Expect five-figure annual contracts based on organization size.

Strengths: If the question you're trying to answer is "do our services meet our standards, and which teams are falling behind," Cortex answers it well. The initiatives feature lets you define campaigns (migrate all services to Node 20, add OpenTelemetry instrumentation) and track adoption over time. Integrations cover most of the tools in a typical backend stack.

Limitations: Cortex is narrower by design. It measures and reports on service maturity, but it doesn't provide self-service workflows, infrastructure management, or CI/CD. If you need a full portal with custom plugins, Cortex won't get you there. The enterprise pricing model also makes it hard to justify for smaller organizations where service sprawl isn't the primary pain.

Best for: Mid-to-large engineering organizations where the main problem is inconsistent engineering standards and lack of visibility into service maturity across teams.

OpsLevel

OpsLevel overlaps with Cortex in scope: service catalog, ownership tracking, and maturity scorecards. The distinguishing feature is checks — automated validations that run against your services and can gate CI/CD pipelines.

Pricing: Per-user pricing, with plans tiered by features. Enterprise plans are custom-priced.

Strengths: Ownership and dependency mapping is well-executed. When an incident happens at 2am, OpsLevel answers "who owns this service" without anyone digging through git history or Slack. The checks system integrates with CI/CD to enforce maturity gates: a service can't deploy to production unless it meets specific criteria. Clean interface with less configuration overhead than Backstage.

Limitations: Similar constraints as Cortex — OpsLevel organizes and measures your services but doesn't manage infrastructure or deployments. The per-service pricing model can get expensive for organizations with hundreds of microservices, and some features that feel core (like certain integrations) are gated behind higher tiers.

Best for: Growing organizations where service ownership has become ambiguous and you need both visibility and automated enforcement of engineering standards.

Humanitec

Humanitec approaches the problem differently from the portals above. It's a platform orchestrator: instead of displaying information about your infrastructure, Humanitec manages how infrastructure gets provisioned through a concept called Score (an open-source workload specification) and dynamic configuration management.

Pricing: Custom enterprise pricing. Score is open source.

Strengths: Humanitec actually provisions infrastructure, which sets it apart from the portal tools. Developers define what they need in a Score file (a database, a DNS record, a Redis cache), and Humanitec resolves those requirements against resource definitions that platform engineers configure. This separation means developers describe intent, and the platform team controls how that intent maps to actual cloud resources. The model works well for organizations that want to give developers self-service without giving them direct cloud access.

Limitations: Humanitec adds a layer of abstraction between developers and infrastructure. The resource definitions that platform engineers write are a new configuration language to learn and maintain. You also still need the underlying infrastructure toolchain — Humanitec orchestrates Terraform, Helm, or whatever provisioning tool you use, rather than replacing it. The orchestration layer can be hard to debug when deployments fail, because the abstraction obscures what's happening underneath.

Best for: Organizations with a platform team that wants to standardize how developers request and receive infrastructure, while keeping control over provisioning details.

Encore Cloud

Encore Cloud is a developer platform that combines a backend framework with infrastructure provisioning, CI/CD, environments, and observability. The approach is different from every other tool on this list: instead of sitting on top of your existing stack or orchestrating existing tools, Encore replaces most of the stack.

You write your backend using the Encore framework (TypeScript or Go), declaring infrastructure needs in application code. new SQLDatabase("orders") provisions PostgreSQL locally and RDS or Cloud SQL in the cloud. new Topic("order-created") provisions NSQ locally and SNS/SQS or GCP Pub/Sub in production. The framework parses these declarations and provisions everything automatically in your own AWS or GCP account.

Pricing: Free tier for hobbyists. Pro at $49/member/month includes production environments, custom domains, and team features. Enterprise pricing is custom.

Strengths: The service catalog, architecture diagrams, and API documentation are generated automatically from code — not configured manually or synced from external tools. CI/CD, preview environments per pull request, distributed tracing, structured logging, and metrics all come built-in. There's no Terraform to write, no Kubernetes to manage, no Datadog to configure. For teams without dedicated platform or DevOps engineers, Encore provides what a fully-staffed platform team would build, without needing to hire or maintain it. The framework also means AI agents write standard application code rather than infrastructure configuration, which eliminates the infrastructure review bottleneck when using AI coding tools.

Limitations: You need to use the Encore framework, which means starting new projects with Encore or migrating existing code. If you have a large codebase on Express or Fastify, that migration is real work. The framework is opinionated about service structure and infrastructure declarations. AWS and GCP are supported, Azure is not. Fine-grained infrastructure control is limited compared to writing Terraform directly — you get the platform's provisioning defaults, which cover common patterns well but may not suit edge cases.

Best for: Teams building new backend projects (or willing to migrate) who want a complete platform without assembling and maintaining separate tools for CI/CD, infrastructure, and observability.

How to Choose: Portal vs Orchestrator vs Platform

The category matters more than the feature list. These tools solve different problems, and the right choice depends on what you're missing.

You need a portal if you already have a working infrastructure and deployment stack but lack visibility. Your CI/CD works, your Terraform modules are maintained, your monitoring is set up — but developers can't find documentation, nobody knows who owns what, and onboarding takes weeks because there's no single place to see the architecture. Backstage, Port, Cortex, and OpsLevel address this. Backstage and Port are general-purpose portals; Cortex and OpsLevel specialize in standards and ownership.

You need an orchestrator if you have the infrastructure toolchain but want to standardize how developers interact with it. Platform engineers manage the complexity; developers get a simplified interface. Humanitec fits here, providing a layer between developer intent and infrastructure provisioning.

You need a platform if you don't have a mature infrastructure toolchain and don't want to build one. Encore Cloud replaces the need for Terraform, Kubernetes, a CI/CD pipeline, an observability stack, and a developer portal by providing all of those as a single integrated system. The trade-off is adopting the framework, which shapes how you write your backend code.

The most common mistake is buying a portal when you need a platform. Adding a service catalog on top of infrastructure you're struggling to manage doesn't make the infrastructure easier to manage — it makes the struggle more visible. If the underlying problem is that your team spends too much time on DevOps and infrastructure, a portal won't fix that. A platform might.

Related Resources

• Internal Developer Platform — what IDPs are and how they work
• Backstage Alternatives — deeper comparison of developer portal options
• Kubernetes Alternatives — container deployment options for smaller teams
• Cloud Infrastructure Automation Tools — comparing Terraform, Pulumi, CDK, and more