BigQuery pricing: $6.25 per TiB on-demand vs editions slot-hours

BigQuery charges for queries in two ways. Under on-demand pricing, Google Cloud bills $6.25 per TiB of data processed, and the first 1 TiB each month is free. Under capacity pricing, a team buys compute through one of three BigQuery editions, Standard, Enterprise and Enterprise Plus, and pays per slot-hour: $0.04, $0.06 and $0.10 per slot-hour respectively at the default pay-as-you-go rates shown on Google Cloud's BigQuery pricing page, checked on November 9, 2026. Storage is billed separately in both models.

The choice between the two decides how a data team's bill behaves. On-demand cost follows the bytes each query reads, so a badly written dashboard query on a large table costs money every time it refreshes. Editions cost follows the slots the autoscaler assigns over time, regardless of how many bytes are scanned, so heavy scanning becomes cheaper while idle-looking reservations can still produce charges. The worked example below shows where the two lines cross for a small analytics team.

Prices on the pricing page vary by location, and commitments are regional. Every figure in the tables that follow is the default list rate as published by Google Cloud; teams should select their own region on the pricing page before budgeting.

··········

BIGQUERY ON-DEMAND PRICING PER TIB.

On-demand is the default model for a new project. Google Cloud lists the price as $6.25 per TiB, with the first 1 TiB of query data processed per month free. The pricing page adds two billing rules that affect small queries: charges are rounded up to the nearest MB, and there is a minimum of 10 MB processed per table referenced by a query and 10 MB per query.

The amount billed depends on the columns and partitions a query reads, not on the rows it returns. Google's cost best-practices page states that for non-clustered tables a LIMIT clause does not affect the amount of data read; on clustered tables a LIMIT can reduce bytes scanned because the scan can stop early. A dry run, or the query validator in the console, reports the estimated bytes before a query runs, and Google describes that estimate as an upper bound on the bytes billed. Teams that write SQL with an assistant can follow the checks in our guide on how to write SQL with Gemini in BigQuery and check the bytes it scans.

Two controls exist to cap on-demand spending. The maximum bytes billed setting makes a query fail without charge when its estimated bytes exceed the limit. Custom quotas limit the data processed per day, per project or per user, and Google's documentation says users cannot run queries once the quota is reached. The same best-practices page describes daily quotas as the only way to restrict costs under on-demand pricing.

··········

BIGQUERY EDITIONS PRICING PER SLOT-HOUR.

Capacity pricing replaces the per-TiB charge with a charge for slots, the units of compute that BigQuery uses to run queries. According to the pricing page, capacity pricing covers query costs, including BigQuery ML, DML and DDL statements, but does not cover storage, BI Engine, streaming inserts or the BigQuery Storage API.

........

Edition	Pay-as-you-go rate	Commitments	Max reservation size
Standard	$0.04 per slot-hour	Not available	1,600 slots
Enterprise	$0.06 per slot-hour	1-year at 20% off, 3-year at 40% off	Quota-based
Enterprise Plus	$0.10 per slot-hour	1-year at 20% off, 3-year at 40% off	Quota-based
On-demand (no edition)	$6.25 per TiB, first 1 TiB free	Not applicable	Quota-based

........

Pay-as-you-go capacity is billed per second with a one-minute minimum by default. Google's autoscaling documentation states that slots always autoscale to a multiple of 50 and that customers are charged for the number of scaled slots, not the number of slots used. Fluid scaling, which can be enabled at the reservation level, removes the one-minute minimum but keeps the 50-slot increments.

Enterprise and Enterprise Plus reservations can also have baseline slots. The autoscaling documentation says baseline slots are always charged, and autoscaled slots are added only after the baseline slots, and idle slots where applicable, are consumed. Commitments are sold for one or three years, with a 50-slot minimum and increments of 50 slots, they renew automatically unless set to cancel, and a commitment bought in one region cannot be used in another.

··········

WHAT EACH BIGQUERY EDITION INCLUDES BEYOND PRICE.

The edition determines more than the slot rate. Google's editions overview lists administration and analysis differences that often settle the decision before cost does.

........

Feature	Standard	Enterprise	Enterprise Plus
Monthly SLO	99.9% or higher	99.99% or higher	99.99% or higher
Compute model	Autoscaling	Autoscaling and baseline	Autoscaling and baseline
BI Engine acceleration	No	Yes	Yes
Create materialized views	No, query existing only	Yes	Yes
Cross-user result caching	No	Yes	Yes
Assured Workloads compliance controls	No	No	Yes
Managed disaster recovery	No	No	Yes

........

Standard is limited to 10 reservations per administration project and up to 16,000 slots per organization, its reservations accept only QUERY and PIPELINE assignments, and it cannot set a maximum concurrency target. Enterprise adds idle capacity sharing and target concurrency. On-demand projects get BI Engine, materialized view creation and refresh, and Assured Workloads compliance controls, but no reservations or advanced workload management. Google also advises against using editions to restrict access to features, because the features in each edition can change over time.

··········

BIGQUERY STORAGE PRICING: LOGICAL VS PHYSICAL.

Storage is charged the same way under on-demand and editions. The pricing page lists rates per GiB-hour for four storage types, with the first 10 GiB per month free for each type. Data that has not been modified for 90 consecutive days moves to the long-term rate, which Google describes as roughly 50% lower.

........

Storage type	Price per GiB-hour	Equivalent per GiB, 730-hour month
Active logical	$0.000031507	$0.023
Long-term logical	$0.000021918	$0.016
Active physical	$0.000054795	$0.040
Long-term physical	$0.000027397	$0.020

........

The monthly column is an illustrative conversion at 730 hours, computed in Python; Google bills the hourly rate. Logical storage measures uncompressed bytes, and Google does not charge for time travel and fail-safe storage under logical billing. Physical storage measures compressed bytes and does charge for time travel and fail-safe, so the lower logical rate does not always produce the lower bill.

Ingestion has its own charges. Batch loading is free when it uses the shared slot pool. Streaming inserts cost $0.01 per 200 MiB, with a 1 KB minimum per row, and the Storage Write API costs $0.025 per GiB, with the first 2 TiB per month free.

··········

WORKED COST EXAMPLE: ON-DEMAND VS STANDARD EDITION.

The following scenario is illustrative and uses the default list rates above. A five-person analytics team runs dashboards and ad hoc queries that process 40 TiB per month. It stores 3 TiB of active logical data and 5 TiB of long-term logical data. The alternative is a Standard edition reservation in which the autoscaler holds an average of 100 slots for 10 hours a day on 22 working days, or 22,000 slot-hours.

........

Cost line (illustrative)	On-demand	Standard edition
Query compute, 40 TiB scanned	$243.75	$880.00
Active logical storage, 3 TiB	$70.43	$70.43
Long-term logical storage, 5 TiB	$81.76	$81.76
Monthly total	$395.94	$1,032.19

........

On-demand compute is 39 billable TiB at $6.25 after the free tier. Standard compute is 22,000 slot-hours at $0.04. Storage subtracts the 10 GiB free tier from each type and multiplies by the hourly rate over 730 hours.

At this volume on-demand costs less than half as much. The picture reverses as scanning grows: the same 22,000 slot-hours cost $880 regardless of bytes, and on-demand reaches that amount at about 141.8 TiB per month. At 300 TiB per month, on-demand compute would cost $1,868.75 against the same $880 for Standard. With Enterprise at $0.06, the same slot-hours cost $1,320, and the break-even rises to about 212.2 TiB per month.

Commitments only pay off for steady, around-the-clock capacity. A 100-slot Enterprise baseline running all 730 hours of a month costs $4,380 at pay-as-you-go rates, $3,504 with a one-year commitment at 20% off and $2,628 with a three-year commitment at 40% off. The team in this example uses slots for about 220 hours a month, so a commitment would cost far more than its autoscaled usage.

··········

COMMON BIGQUERY PRICING MISUNDERSTANDINGS.

Several assumptions that analysts bring to BigQuery are contradicted by Google's own documentation. A LIMIT clause does not lower the on-demand cost of a query on a non-clustered table, because the whole set of referenced columns is still read. An autoscaling reservation is billed on the slots it scales to, in multiples of 50, so a query that needs 60 slots is billed at 100. Capacity pricing does not cover storage, BI Engine or streaming inserts, which continue to appear as separate lines on the invoice.

Standard edition cannot buy commitments, and its reservations are capped at 1,600 slots, so a team that needs discounted long-term capacity or larger reservations has to start at Enterprise. Pricing changes also reach specific features over time: Google recently extended on-demand pricing to graph queries, as covered in our article on BigQuery Graph running GQL queries on on-demand pricing.

Teams comparing warehouses can set these figures against the per-credit model described in our reference on Snowflake pricing, credits, editions and warehouse sizes. The units differ: BigQuery on-demand bills data scanned, while both BigQuery editions and Snowflake warehouses bill compute time.

··········

BIGQUERY ON-DEMAND OR EDITIONS: WHAT TO CHECK BEFORE CHOOSING.

The list rates are confirmed by Google Cloud: $6.25 per TiB on-demand with 1 TiB free each month, and $0.04, $0.06 and $0.10 per slot-hour for Standard, Enterprise and Enterprise Plus at pay-as-you-go, with one-year and three-year commitments at 20% and 40% off for the two higher editions. Rates vary by location, so the first check is the region of the datasets in the pricing page's location selector.

The second check is the team's real scan volume. The console's job history and the dry-run estimates show bytes processed per query, and Google offers on-demand customers sizing recommendations for moving to capacity pricing. A team scanning well under the break-even volume for its edition generally pays less on-demand, provided it sets maximum bytes billed and custom daily quotas.

The third check is features rather than price. BI Engine, materialized view creation, cross-user caching, compliance controls and managed disaster recovery are tied to specific editions, and a requirement for any of them can make Enterprise or Enterprise Plus necessary even when Standard slot-hours would be cheaper. Negotiated discounts and contract terms are not published and have to be confirmed with Google Cloud sales.

··········

FOLLOW US FOR MORE.

·····

DATA STUDIOS

·····

datastudios.org