BigQuery Cost per TB: Exact Rates, Storage Tiers, and What You’re Actually Paying

By Amir Peres
July 30, 2026 | 5 min read
Illustration of a cloud data warehouse with tiered storage layers and stacks of coins representing BigQuery cost per TB

Most BigQuery cost questions start like this: someone pulls up the pricing page, sees “$6.25 per TiB,” and assumes that’s the number to build a budget around. Then the invoice arrives… and finance is not happy.

The truth about the “$6.25 per TiB” is that you’re only looking at the on-demand query rate. That number says nothing about:

  • What your data type choices cost in storage
  • How physical versus logical billing affects that number
  • What BigQuery ML training adds on top
  • How the same terabyte of data costs a completely different amount depending on where it’s stored and how old it is

This article breaks down BigQuery cost per TB across every dimension. We’ll cover all you need to know about query processing, storage tiers, data types, and additional services. By the time you finish reading this, you’ll understand exactly what’s on your bill and why.

Note: All figures reflect U.S. multi-region rates from Google Cloud’s official pricing documentation. Pricing may vary from region to region. Always verify current rates.

BigQuery Query Cost Per TB: The On-Demand Rate

The most commonly cited figure – and the one that matters most for teams on the default pricing model – is the on-demand query rate of $6.25 per TiB scanned.

But let’s break this down a bit more:

  • Per GiB: $0.00610 (or approximately $0.006)
  • Per GB: $0.00625
  • Free tier: The first 1 TiB of query data processed per month is free per billing account

BigQuery charges based on bytes scanned by the query, not bytes returned. A query that touches 1 TiB of column data but returns 10 rows still costs $6.25 minus the free tier.

Note: There’s also a minimum charge. Every table referenced in a query is billed for at least 10 MB of data processed, even if the actual scan is smaller. This matters for queries that touch many small lookup tables.

Here’s a quick reference for what on-demand query costs look like at scale:

Data ProcessedCost (after free tier)
1 GiB~$0.006
10 GiB~$0.061
100 GiB~$0.61
500 GiB~$3.05
1 TiB$6.25
10 TiB$62.50
100 TiB$625.00

Note: These prices assume U.S. multi-region on-demand rate of $6.25 per TiB, after the first 1 TiB free.

Notice how the jump from 10 TiB to 100 TiB is linear. There are no volume discounts for on-demand pricing. This is one of the many reasons high-volume teams move to capacity pricing because the math works out better at scale.

BigQuery Storage Cost Per TB

Storage in BigQuery is billed independently from compute and across two dimensions: how old the data is (active vs. long-term) and how it’s measured (logical vs. physical). We’ll break down how costs work per TB for each in more detail below.

Active vs. Long-Term Storage

Any table or partition modified in the last 90 days is considered and billed as active storage.

Data that is not modified for 90 days is considered long-term storage – and is charged at half the rate.

No manual action is required here. BigQuery handles the transition automatically, and there is no performance or availability difference between the two tiers.

Note: “Modified” means any write operation (loading data, copying, streaming inserts, or DML statements like INSERT, UPDATE, DELETE. Running queries against a table does not count as modification and will not reset the 90 day clock.

Logical vs. Physical Storage Billing

This is the choice most teams should put more time into considering. Logical billing charges based on the uncompressed size of your data. Physical billing charges based on the actual bytes on the disk after compression.

Because BigQuery uses columnar storage with efficient compressions, physical bytes are often a small fraction of logical bytes. This is especially the case for structured, repetitive data like event logs or transactional records.

Here’s what the cost breakdown looks like for active vs. long-term storage and logical vs. physical storage billing at the query level:

Storage TypeCost Per GiB/MonthCost Per TB/MonthFree Tier
Active Logical$0.02$20.00First 10 GiB/month
Active Physical$0.04$40.00First 10 GiB/month
Long-Term Logical$0.01$10.00First 10 GiB/month
Long-Term Physical$0.02$20.00First 10 GiB/month

Note: This pricing uses U.S. multi-region rates. Physical billing invoices time travel and fail-safe storage separately at active physical rates – these are not included in the table above.

At first glance, physical billing appears more expensive per GiB. But that’s the rate on compressed bytes.

If your data compresses at 3:1, your effective cost under physical billing is $0.04 divided by 3, or roughly $0.013 per logical GiB, which is much cheaper than the $0.02 logical rate. At 5:1 compression, the effective rate drops to $0.008.

Note: Storage costs vary by region. Some regional prices have quite large gaps. For reference:

  • Mumbai (Asia-south1): ~$0.023 per GiB for active logical storage
  • Sydney (Australia-southeast1): ~$0.025 per GiB
  • Singapore (Asia-southeast1) and Tokyo (Asia-northeast1): ~$0.022 per GiB

Teams with multi-region data strategies should account for this with budget monitoring.

BigQuery Storage Cost by Data Type

BigQuery charges for storage based on the actual byte size of each column’s data type. This matters for schema design decisions at scale.

Here are the byte sizes per value for BigQuery’s core data types:

Data TypeBytes Per Value
INT64 / INTEGER8 bytes
FLOAT64 / FLOAT8 bytes
NUMERIC16 bytes
BIGNUMERIC32 bytes
BOOL / BOOLEAN1 byte
STRING2 bytes + length of string
BYTES2 bytes + number of bytes
DATE8 bytes
DATETIME8 bytes
TIME8 bytes
TIMESTAMP8 bytes
GEOGRAPHY16 bytes + size of encoding
JSONSize of UTF-8 encoding
ARRAY0 bytes + size of elements
STRUCT / RECORD0 bytes + size of fields
NULL values0 bytes

BigQuery ML Cost Per TB

BigQuery ML pricing is different from standard query processing. It depends on the model type and your pricing model (on-demand vs. Editions).

For on-demand users, BigQuery ML training costs are billed per TiB of data processed – but at a much higher rate than standard queries for many workloads.

  • BigQuery ML model training (built-in models): ~$312.50 per TiB for training operations
  • BigQuery ML prediction/evaluation: Billed the same as standard on-demand queries at $6.25

Built-in models trained within BigQuery include linear regression, logistic regression, k-means, matrix factorization, PCA, and ARIMA. External models (DNN, boosted tree, random forest, AutoML) are trained on Vertex AI and carry Vertex AI pricing in addition to BigQuery ML costs.

If your team uses Edition, BigQuery ML is included within the Enterprise and Enterprise Plus tiers – so you’ll pay for slots consumed during training instead of per TiB.

What does this mean practically? If you’re training ML models in BigQuery on-demand, the per-TiB rate is 50x the standard query rate. So that leaves you with workloads that were once manageable at $6.25/TiB that are now driving up prices. This is why teams that do significant ML training often prefer the Editions model.

Streaming Insert Cost Per TB

Streaming inserts – real-time data ingestion via BigQuery streaming APIs – are billed separately from storage and query computation.

The streaming insert rate is $0.01 per 200 MB, which equals $0.05 per GB, or roughly $51.20 per TiB.

That’s more than 8x the standard on-demand query rate on a per-TiB basis.

Pro tip: Batch loading data (from Cloud Storage, local files, or most Google services), is free. Batch loading is often more cost-effective if your pipeline does not require immediate data availability.

BigQuery Free Tier: What’s Actually Free Each Month

BigQuery does have a free tier. And it is worth taking advantage of. Here’s what BigQuery gives you free of charge each month:

  • Query processing: First 1 TiB per month (per billing account)
  • Storage: First 10 GiB per month (per billing account)
  • Data loading: Batch loads from Cloud Storage and local files
  • Data export: To Cloud Storage within the same region
  • Metadata operations: Table creation, deletion, schema updates

What isn’t free?

  • Streaming inserts
  • Cross-region data transfers
  • Storage Read API reads
  • BigQuery ML training

How to Manage Per TB Cost and Under Capacity Pricing

Teams that are processing consistent, high volumes find that BigQuery Editions (capacity pricing) replaces the per-TiB model entirely because you can pay for slots instead of per byte scanned. The per-TiB framing stops applying, but you can estimate an effective per-TiB rate for comparison.

Here’s how that math breaks down:

At the Standard Edition pay-as-you-go rate of $0.04 per slot-hour. At 100 slots running for a full month, that will cost you $2,920.

If those slots process:

  • 1,000 TiB per month, your effective cost is $2.92 per TiB – less than half the on-demand rate.
  • At 500 TiB, the effective cost is $5.84, which is nearly the same as on-demand.
  • At 200 TiB, it’s $14.60 per TiB, considerably more expensive than on-demand.

This shows you why slot reservations are a strong choice for high-volume, steady workloads. Unpredictable workloads often lead you to overpaying because you’re buying capacity that just sits idle.

Why BigQuery Per TB Costs Are Just the Beginning

The $6.25 per TiB number is a good starting point, but actual BigQuery spend is shaped by decisions that happen before a query even runs: schema design, storage billing model, data lifecycle, pipeline architecture, and ML workloads all play a role in your final bill.

Teams that understand that full cost picture are the ones who can accurately forecast and optimize spend. Teams that don’t? They’re the ones left concerned with how much each monthly bill is going to cost them.

If your BigQuery invoice is growing faster than your data volume – or you’re not certain what’s driving up those costs – Yuki’s BigQuery cost optimization platform can help shed some light. Our tool brings clarity to:

  • Query-level attribution
  • Anomaly detection
  • Continuous optimization

All without requiring changes to your existing pipeline – or any additional dev lift.

Curious to see how much work your BigQuery investments can do for you when fully optimized? Get your free Yuki demo today.

By Amir Peres
Amir Peres is CTO and Co-Founder of Yuki, where he drives technical vision for automated Snowflake cost optimization. With 12+ years in data architecture, ML, and large-scale infrastructure, he previously led engineering at Lightico (building GDPR-compliant multi-region data lakes) and Payoneer (ML product development). Amir specializes in scalable, secure, cost-efficient data systems that maximize ROI while reducing manual effort. He has presented at Data TLV Summit 2025 and appeared on the Jon Myer podcast. Find more of his insights on LinkedIn.

Table of Contents

Free cost analysis

Take 5 minutes to learn how much money you can save on your Snowflake account.

By clicking Submit you’re confirming that you agree with our Terms and Conditions.

Follow us on LinkedIn

Related posts

Free cost analysis

Take 5 minutes to learn how much money you can save on your Snowflake account.

By clicking Submit you’re confirming that you agree with our Terms and Conditions.

Skip to content