Astro review

Astro is Astronomer’s platform for running Apache Airflow data and AI workflows in production. It combines an Airflow runtime with tools for deploying, developing and monitoring workflows; it is not a replacement for Airflow’s DAG-based approach. Astro also includes Otto, an AI agent described as able to write DAGs, investigate failures and plan upgrades using team context and private memory.

Teams can develop locally with the Astro CLI or in a browser-based IDE. Deployment settings can be managed through Git-reviewed configuration, the CLI, API or Terraform. For scheduling, Astro supports event-driven starts when new work arrives, rather than waiting for a polling cycle. Its observability tools cover health, data quality, SLAs and lineage, and the platform supports rolling back to deployments from the last 90 days. In short, Astro adds a managed operating layer around Airflow, with a price tied to deployment and worker usage.

Astro keeps Airflow at the center, which suits teams that want Airflow workflows but need a platform for production deployments and operations. Its feature set covers the work around DAGs: local or browser-based development, configuration managed through Git, CLI, API or Terraform, and monitoring for health, data quality, SLAs and lineage.

The platform’s event-driven scheduling can start tasks as work arrives instead of waiting for a polling cycle. Astro also lists deployment rollbacks to versions from the last 90 days. Otto may help teams draft DAGs, investigate failures or plan upgrades; its described use of team context and private memory is aimed at work within a team’s existing Airflow environment.

The Developer tier lists deployments starting at $0.35 per hour, with scale-to-zero compute, hibernating deployments, API access and rollbacks. Team starts at $0.42 per deployment per hour and adds dedicated clusters, high availability, 7-day audit-log retention and 24x5 support. Those published starting rates give buyers a concrete entry point, even though they are not monthly plan totals.

Astro’s pricing is usage-based, so the hourly starting rates do not tell you what a deployment will cost over a month or year. The pricing page shows monthly and annual options but does not publish totals. Developer starts at $0.35 per deployment per hour and Team at $0.42; dedicated single-tenant clusters start at $2.40 per hour, and workers start at $0.13 per hour. Cloud-provider networking costs are passed through.

Several costs remain open-ended: Astro Observe is priced as a percentage of deployment and worker compute, but the rate is not published. AI token usage is extra, and its rate is also not published. Business, Enterprise, Enterprise Business Critical and Astro Private Cloud require custom quotes. The pricing information lists no free plan, and a free-trial term is not published.

There are also tier-specific limits to weigh: Team lists 7-day audit-log retention and 24x5 support, while Business lists 90-day retention and 24x7 support. Buyers needing longer retention, round-the-clock support or the listed one-hour and 30-minute SLAs must evaluate higher, custom-priced tiers.

Astro does not publish a fixed monthly total in the published pricing details. The Developer tier starts at $0.35 per deployment per hour; Team starts at $0.42. That is a $0.07-per-hour increase per deployment, before add-ons or other usage. The pricing page shows monthly and annual options, but lists hourly starting rates rather than monthly or annual totals.

Developer includes scale-to-zero compute, hibernating deployments, API access and rollbacks. Team adds dedicated clusters, high availability, seven-day audit-log retention and 24x5 support. Business and higher tiers require custom quotes; Business lists 90-day audit logs, 24x7 support and a one-hour SLA.

Budget for extras: dedicated single-tenant clusters start at $2.40 per hour, workers at $0.13 per hour, and cloud networking costs pass through. Astro Observe is priced as a percentage of deployment and worker compute, while AI token rates are not published. If you need a predictable bill, those unknowns matter.

Astro runs Apache Airflow, so it is a more natural move for teams already building Airflow workflows than for teams changing orchestration systems. Development is available through the Astro CLI or browser-based Astro IDE, and deployments can be managed through Git-reviewed configuration, CLI, API or Terraform. The published information does not detail DAG migration steps, data export, or an exit process, so confirm those before moving production workloads.

Astro is not a fit if you do not want Airflow, and its hourly deployment pricing plus unpublished AI and Observe rates may be reasons to look elsewhere. But we have no Reddit posts or comments in the published material to verify what users actually cite when leaving. Do not treat those fit concerns as reported churn reasons. Ask Astro how to retrieve workflow definitions, logs and configuration if you leave, and whether any Astro-specific setup must be replaced. The available details do not establish how portable those items are.