BigQuery pricing, made clear.
Explore the options. See what changes your bill. Find your next move.
or slots from $0.04 / slot-hour
Active: ≈ $0.023 or $0.040 / GiB-month
Memory for eligible accelerated queries.
Step 1 of 4: Compute
Pay for data read. Or for capacity.
Explore each setup. BI Engine adds acceleration; reservations change how compute is billed.
The simplest way to use BigQuery
Start querying without sizing a reservation. Pricing is straightforward: pay for the billable data your queries read.
Clear unit pricing makes costs easy to estimate. Reading less data reduces on-demand query charges.
(100 − 1 free TiB) × $6.25
In Biq Blue’s field experience, BI Engine generally reduces on-demand query charges by around 20%. Accelerated work is not charged as on-demand reads. BI Engine memory is billed separately.
Fluid Scaling
Autoscaled capacity adjusts second by second, with no one-minute minimum.
Default autoscaling retains this burst for the 60-second minimum.
Simplified example: constant 100-slot allocation, no baseline or other queries. Monthly savings depend on your allocation pattern; the monthly estimate above uses the slot-hours you enter.
On-demand is the simplest starting point. Other projects can use reservations when their workloads benefit from capacity pricing.
Each setup uses different example inputs. Comparing the same workload requires measured usage. Storage and other services are extra.
What else changes the comparison?
Reservations can autoscale; commitments are optional. Compare billed allocated slot-hours, not just query execution time. BI Engine also works with Enterprise and Enterprise Plus reservations, but not Standard. Some edition commitments include BI memory. These examples use list prices without commitment or bundled-memory discounts. Edition features ↗
What if your query could skip most of the table?
Turn on each optimization. Watch the blocks read by this example query.
150 of 150 blocks read
The query reads everything to find the rows it needs.
With on-demand, fewer billed bytes can lower query charges. With reservations, less work helps efficiency; the bill falls only if billed capacity decreases.
Illustration: 30 days × 100 GiB, with 5 equal customer blocks per day. Free query allowance already used. Actual pruning depends on your filters, data and clustering order.
What makes pruning work in real queries?
Use a filter that allows BigQuery to eliminate partitions, and filters on useful clustering columns. Partitioning and clustering can work together. Selecting only the columns you need also reduces reads; adding LIMIT to SELECT * does not by itself reduce scan charges. Partition pruning ↗ · Clustering ↗
Same data. Two ways to measure it.
Compression helps physical billing. Retained history adds to it. Move the sliders to see the trade-off.
Logical
Bill the uncompressed data size.
History included in the base rate.
Physical
Bill compressed data + retained history.
Active-storage example, before the free tier; rounded monthly rates based on 730 hours. Changing the billing model does not recompress or move your data. Choose per dataset.
I don’t know my compression or history sizes
That is normal. Compare logical bytes, current compressed bytes and retained historical bytes from BigQuery’s storage metadata. When using ACTIVE_PHYSICAL_BYTES, time travel is already included; do not count it twice. Add fail-safe separately. Find the storage measurements ↗
What about long-term storage?
Eligible data that has not been modified for 90 consecutive days moves to long-term pricing. Model active and long-term bytes separately.
| USD / GiB-month | Active | Long-term |
|---|---|---|
| Logical | ≈ $0.023 | ≈ $0.016 |
| Physical | ≈ $0.040 | ≈ $0.020 |
Now you know what to compare.
Here are the checks that turn a price list into a decision for your BigQuery.
- 01
Compare compute on real demand
Start with straightforward on-demand pricing. Compare reservations for heavy pipelines, and check whether Fluid Scaling could reduce billed capacity.
Measure: billed TiB · allocated slot-hours · concurrency · autoscaling pattern - 02
Accelerate dashboards with BI Engine
For repeated dashboard queries, compare the reduction in on-demand charges with the cost of reserved BI memory.
Measure: query eligibility · memory needed · remaining billed bytes - 03
Find data your queries could skip
Check partition filters, clustering columns and unnecessarily selected columns. These optimizations can help under either compute model.
Inspect: query filters · table layout · bytes billed - 04
Compare both storage meters
Use each dataset’s compression, active/long-term data and retained history. The cheaper option can differ by dataset.
Measure: logical bytes · compressed bytes · time travel · fail-safe
Discover what you could save before changing anything.
Biq Blue automatically analyzes your BigQuery usage and estimates the savings from pricing and configuration changes. See the opportunities first, then decide what to apply.
Discover my savingsAutomatic analysis of usage metadata. You decide which changes to make.
Automatic analysis of your usage, before you change anything.