AWS Glue pricing guide details
Pricing last verified: April 2026. Plans and pricing may change — check the vendor site for current details.
Pricing Overview
AWS Glue uses a fully usage-based pricing model with no upfront costs or monthly minimums. The first million objects stored are free, and the first million accesses are free. The price of 1 DPU-hour is $0.44. This serverless model means you pay nothing when jobs are idle — there are no always-on clusters to maintain. For organizations already running on AWS, Glue integrates natively with S3, Redshift, Athena, and other services without cross-platform data transfer fees within the same region.
Plan Comparison
| Component | Price | What It Covers |
|---|---|---|
| Data Catalog free allowance | $0 | The first million metadata objects stored and the first million accesses |
| Data Catalog metadata above the free allowance | $1.00 per 100,000 objects per month | Metadata objects over 1 million |
| ETL jobs | $0.44 per DPU-hour | Processing and loading data; billed by the second |
| Interactive Sessions | $0.44 per DPU-hour | Interactive development of ETL code; an Interactive Session has 5 DPU by default |
| DataBrew interactive sessions | Billed per session | Interactive sessions in AWS Glue DataBrew |
| DataBrew jobs | Billed per minute | AWS Glue DataBrew jobs |
AWS Glue pricing can vary by AWS Region. For an AWS Glue Apache Spark job that runs for 15 minutes and uses 6 DPU, the example price is $0.66: 6 DPU × 0.25 hour × $0.44 per DPU-hour.
Hidden Costs and Considerations
AWS Glue charges an hourly rate, billed by the second, for crawlers and ETL jobs. In AWS's ETL pricing example, the price is $0.44 per DPU-hour, so job duration and the number of DPUs used directly affect processing costs. Pricing can vary by AWS Region.
Data Catalog costs can also apply when usage exceeds the free tier: the first million stored metadata objects and the first million accesses are free. Additional charges may arise from related AWS services. For example, data stored in Amazon S3 or Amazon Redshift is charged at those services' standard storage rates, and Amazon S3 storage, requests, and data transfer can apply when Glue processes data from S3.
Cost Estimates by Team Size
Glue has no per-seat charge, so team size only matters through how much data the team moves. Every estimate below builds on the published rate of $0.44 per DPU-hour, billed by the second.
- Small team (2-3 engineers, a few daily jobs): a Spark job on 6 DPUs that runs 15 minutes costs $0.66. Twenty such runs a day is roughly 60 DPU-hours a month. Crawlers bill at the same DPU rate, and the first million Data Catalog objects and the first million accesses are free, so most small teams stay inside the free catalog tier.
- Mid-size team (10-20 engineers, hourly pipelines): interactive development becomes the second line item. A Glue Studio notebook session runs on 5 DPUs by default, so 24 minutes of exploration costs $0.88. Several engineers developing daily adds tens of DPU-hours a month on top of scheduled jobs.
- Large org (terabyte-scale, continuous ingestion): job DPU-hours dominate and the Data Catalog moves past its free tier into the monthly metadata fee. Region matters at this size — AWS states that Glue pricing varies by Region, so read the rate card for the Region the pipelines run in.
Because billing is per second, the levers are job duration and DPU count, not headcount. Cutting a job from 6 DPUs to 4, or from 15 minutes to 10, changes the bill directly; adding engineers does not.
How AWS Glue Pricing Compares
AWS Glue's usage-based model differs fundamentally from the seat-based pricing of most competitors. Dagster charges $29 per month for Pro with per-user scaling, meaning a 10-person team pays $290 per month regardless of data volume. dbt Cloud starts at $25 per month for Pro and $100 per month for Team plans. Hevo Data offers a free tier for 1 million rows and Pro at $25 per month for 10 million rows. For small data volumes, competitors are cheaper — a team processing under 50 GB monthly pays less with Dagster or dbt. For large-scale pipelines processing terabytes, Glue's per-GB model can be more cost-effective since there are no per-seat charges and the serverless model eliminates infrastructure management costs.