At a glance
| Source connectors | No built-in connectors; flows call libraries, integrations or ingestion tools |
| Destination connectors | Same: tasks write through Python libraries and integrations |
| Batch or streaming | Scheduled and event-triggered runs; not a streaming engine |
| Change data capture | Not built in; orchestrate a CDC or ingestion tool |
| Transformations | Any Python or SQL your tasks run, including dbt |
| Orchestration | Core purpose: scheduling, retries, automations and observability |
| Deployment | Self-hosted server or Prefect Cloud; flows run on your infrastructure or Prefect Managed |
| Licensing | Apache License 2.0; Prefect Cloud by plan |
Prefect suits Python teams that want to add scheduling, retries and observability to existing scripts without learning a separate DAG language. The open source framework and server are free, and Prefect Cloud has a permanent free Hobby plan, which makes it easy to start; SSO, RBAC and multiple workspaces are Enterprise features.
Overview
Prefect is a workflow orchestration framework for building data pipelines in Python. You decorate functions with @flow and @task, and Prefect tracks their states, retries failures, caches expensive results and can resume interrupted runs. Flows can be triggered by schedules, external events or API calls, and tasks and branches can be created at run time. The current major version is Prefect 3; the latest release on GitHub when checked was 3.9.2 (6 November 2026), and Prefect 3 requires Python 3.11 or newer.
Every Prefect deployment talks to an API server: either Prefect Cloud or a self-hosted Prefect server you run and scale yourself. Work pools connect deployments to infrastructure: hybrid pools need a worker in your environment, push pools submit runs straight to serverless services such as Google Cloud Run, AWS ECS, Azure Container Instances or Modal, and Prefect Managed pools run flows on Prefect-managed infrastructure in Prefect Cloud.
Key features
- Flows and tasks written as ordinary Python functions with decorators, including async support
- State tracking, retries, caching and resuming runs from the last successful point
- Event-driven automations and triggers, open sourced in Prefect 3
- Work pools for process, Docker, Kubernetes, ECS, Cloud Run and Azure Container Instances
- Push and Prefect Managed work pools that run flows without a long-running worker
- Interactive workflows that pause for approval or input
- Read-only Prefect MCP server for AI assistants to inspect deployments and runs
Pros and cons
Pros
- Little new syntax to learn for Python developers
- Free, Apache-licensed framework and self-hostable server
- Permanent free Hobby plan on Prefect Cloud
- Flexible execution through work pools, including serverless push pools
Cons
- Current releases require Python 3.11 or later
- SSO, RBAC and multiple workspaces are Enterprise features on Prefect Cloud
- Fewer cloud-provider managed offerings than Apache Airflow
- A self-hosted server must be at least as new as the clients that connect to it
Pricing
Prefect and its self-hosted server are free under the Apache License 2.0. Prefect Cloud plans, as listed on prefect.io/pricing on 7 November 2026 (USD, no region stated): Hobby is free (2 users, 5 deployments, 500 minutes a month of serverless compute, 1 workspace, 7-day run retention); Starter is USD 100 a month (3 users, 20 deployments, 75 hours of serverless compute); Team is USD 100 per user a month (4 to 8 users, 100 deployments, 225 hours); Enterprise is custom priced, with SSO, RBAC with object-level permissions, multiple workspaces, SCIM and a 99.9% uptime SLA.
Pricing checked on the vendor's official page on 7 November 2026. Prices change; confirm before buying.
Who it is best for
Python teams that want lightweight orchestration around existing code, with an easy free start on Prefect Cloud.
Who should look elsewhere
- Teams that want a managed orchestrator from their cloud provider: see Apache Airflow vs Prefect.
- Teams that want lineage and data assets at the centre of the tool: see Prefect vs Dagster.
Verdict
Prefect is a practical orchestrator for Python-first teams: it wraps existing functions with retries, caching and automations, and its free Hobby plan lowers the cost of trying Prefect Cloud. Larger organisations should check which governance features sit in Enterprise, and teams that need cloud-provider managed hosting may still lean towards Airflow.
Frequently asked questions
Is Prefect free?
The Prefect framework and self-hosted server are free under the Apache License 2.0. Prefect Cloud has a free Hobby plan (2 users, 5 deployments, 500 serverless minutes a month as listed on 7 November 2026) and paid Starter, Team and Enterprise plans.
Which Python version does Prefect need?
Prefect's installation guide says Prefect 3 requires Python 3.11 or newer.
Do I need a worker to run Prefect flows?
Not always. Hybrid work pools need a worker in your infrastructure, but push pools (Cloud Run, ECS, Azure Container Instances, Modal) and Prefect Managed pools do not.
Sources
- Prefect Cloud pricing
- Prefect installation guide
- Prefect work pools
- Prefect integrations
- Prefect repository (Apache 2.0)
- Prefect releases
Checked November 2026.
How we research reviews and what star ratings mean: our editorial method.