Data APIs

A charge for bulk-reading table data through the BigQuery Storage Read API rather than by running a query. It surprises teams who think of BigQuery as a query engine, because the tools generating it (Spark, Dataflow, BI extracts, client libraries) do not look like queries from the outside. The first 300 TiB a month are free, so a charge here means very large or very repetitive reads.

Billed: Per tebibyte read through the Storage Read API, after a monthly free allowance of 300 TiB per billing account. Separate from query analysis charges: reading via the API is not a query.

A charge for rows written with the legacy streaming API, tabledata.insertAll, which Google now calls the Storage Write API (REST). It is billed per 200 MiB inserted, with every row counted as at least 1 KB, so many tiny rows cost more than their size suggests. Batch loads are free, so this charge is a payment for immediacy.

Billed: Per 200 MiB of rows successfully inserted, with each row counted as at least 1 KB. No monthly free allowance. Batch loading the same data uses a free shared slot pool.

A charge for data ingested through the BigQuery Storage Write API (gRPC), the recommended streaming path. It is billed per gibibyte after 2 TiB a month free, whichever stream type writes it. The stream type changes delivery semantics, not the price.

Billed: Per gibibyte ingested through the Storage Write API (gRPC), after the first 2 TiB each month per billing account, which are free. Every stream type bills the same way.