Cloudflare Basin brings open data pipelines to general availability
Cloudflare Basin combines data ingestion, Iceberg table management and SQL analytics. Its usage pricing separates query scans from the rest of the pipeline.
Cloudflare made Basin generally available on October 1, bringing its data ingestion, catalog and SQL products under one name. The platform stores analytical data on R2 using Apache Iceberg.
Cloudflare connects the launch to the growth of applications built with coding agents. It says developers and agents can query data through an immediately available API without first provisioning a query cluster.
For teams building AI applications, the practical question is how much work remains between collecting an event and getting a useful answer from it. Basin combines several steps, while leaving room to use external analysis tools.
Three services share the data layer
The release notice describes Basin Pipelines as the entry point. It accepts events from HTTP endpoints, Workers and Cloudflare Logpush, applies SQL transformations, and delivers the result as Iceberg tables or files on R2. Typed bindings can help catch schema mistakes, while the dashboard exposes dropped events.
Basin Catalog manages the tables through the standard Iceberg REST interface. Engines including Spark, Snowflake and PyIceberg can use that catalog. R2 does not charge egress fees, which can make working across clouds less expensive, although other services still have their own costs.
Basin SQL supplies Cloudflare’s distributed query engine for the same tables. It uses file pruning and distributed compute to read analytical datasets, and can be called through Wrangler.
The names are changing from Cloudflare Pipelines, R2 Data Catalog and R2 SQL. Cloudflare says existing configurations continue to work. Its changelog also flags deprecation over time without providing a timetable or explaining exactly which legacy elements will be retired.
The scan price is one part of the bill
Basin SQL pricing lists $0.0025 per GB of compressed data scanned, equivalent to $2.50 per TB. It includes 10 GB each month and applies a minimum billable scan of 10 MB per query. R2 storage, operations and catalog charges are additional.
Free ingress does not make the whole pipeline free. The Pipelines rate card charges for SQL transformations and delivery to a destination. Workers Paid includes 50 GB monthly for each dimension. Additional transformations cost $0.04 per GB. Delivery costs $0.03 per GB for JSON or $0.06 for Parquet and Iceberg, measured before compression.
Catalog pricing includes one million operations monthly, then charges $9 per million. Optional compaction adds separate charges for the amount of data and number of objects processed. Those maintenance costs also sit alongside R2 storage and operations.
The SQL scan rate is not a new price cut. Cloudflare’s earlier pricing announcement, dated May 28, already listed $2.50 per TB. October’s development is general availability and the unified Basin product family.
What to check before using it
Cloudflare still describes full SQL data definition support, Iceberg V3 support and more granular catalog authorization as work ahead. Teams should check whether their required operations and access controls are already available.
A useful trial should track a complete workload. Count transformed and delivered bytes, query scans, catalog activity and retained storage. Test the queries the application actually runs and confirm that the chosen external engine can read the tables as expected.
ByteForward has not tested Basin’s performance or reproduced its cost examples. The public documentation supports the product and billing description, while the value for an individual application depends on its workload.
For the model side of Cloudflare’s agent tooling, see ByteForward’s coverage of Clef decision models. Basin addresses the data infrastructure those kinds of applications need.
Featured image is an original AI generated conceptual illustration of shared data infrastructure and an AI agent.
