Skip to content

Compare

DuckHouse vs MotherDuck.
Two ways to run DuckDB in the cloud.

MotherDuck is a managed DuckDB service with named instance sizes, a free Lite plan, and a Business plan at $250 a month plus usage. DuckHouse runs each database as its own DuckDB server, billed by the second for the size you pick, with no plan fee.

MotherDuck’s list prices for its Tier 1 regions (AWS us-east-1 and us-west-2), checked on November 1, 2026. Sources

Price list

Side by side.

Each service’s unit prices, in its own units. DuckHouse’s come from the code that bills it.

DuckHouse and MotherDuck prices
TopicDuckHouseMotherDuck
Smallest computeshared-cpu-1x, 512 MB: $0.0117 an hourPulse: $0.60 per CU-hour, metered by the compute each query uses
Dedicated compute$0.127 to $1.02 an hour, 2 GB to 16 GBStandard $2.40, Jumbo $4.80, Mega $12.00, Giga $24.00 an hour (Business)
Storage$0.02 per GB-month in object storage (DuckLake), $0.225 for a DuckDB disk$0.04 per GB-month, plus 7 days of fail-safe data
Plan feeNone$0 on Lite, $250 a month on Business
Free allowanceNoneLite: 10 GB of storage and 10 CU-hours of Pulse a month

Worked examples

What a month costs.

The same workload on both sides, line by line, so you can check the arithmetic.

Four hours a working day

100 GB of data, queried on a dedicated instance for 4 hours each working day: 88 hours a month.

DuckHouse, performance-8x

performance-8x (8 dedicated vCPUs, 16 GB), 88 hours × $1.0192
$89.69
100 GB in object storage × $0.02
$2.00
A month$91.69

MotherDuck Business, Standard instance

Business plan
$250.00
Standard instance, 88 hours × $2.40
$211.20
100 GB storage × $0.04
$4.00
A month$465.20

MotherDuck doesn’t publish the hardware behind a Standard instance, so DuckHouse is priced at its largest size here, 8 dedicated vCPUs and 16 GB. A smaller size would cost less.

Always on

The same 100 GB, with the instance running all month: 730 hours.

DuckHouse, performance-8x

performance-8x (8 dedicated vCPUs, 16 GB), on all month
$744.00
100 GB in object storage × $0.02
$2.00
A month$746.00

MotherDuck Business, Standard instance

Business plan
$250.00
Standard instance, 730 hours × $2.40
$1,752.00
100 GB storage × $0.04
$4.00
A month$2,006.00

A database that answers an app or a dashboard all day looks like this. On either service, a size that stops when idle costs less.

Small and occasional

5 GB of data and a few queries a week, about 20 hours of running time a month.

DuckHouse, shared-cpu-1x

shared-cpu-1x (1 shared vCPU, 512 MB), 20 hours × $0.0117
$0.23
5 GB in object storage × $0.02
$0.10
A month$0.33

MotherDuck Lite

Lite plan, within 10 GB and 10 CU-hours of Pulse
$0.00
A month$0.00

MotherDuck is cheaper here. Lite costs nothing while you stay within its allowance; its CU-hours count the compute your queries use, not the hours an instance is up, so they aren’t the same unit as DuckHouse’s hours. DuckHouse has no free tier.

Beyond price

Other differences.

Other differences between DuckHouse and MotherDuck
TopicDuckHouseMotherDuck
HardwarePublished for every size: 1 to 8 vCPUs, 512 MB to 16 GBNot published; sizes are named Pulse to Giga
When idleStops when nothing is connected, if you turn on stop when idle; waking takes about five secondsStays up for a cooldown after the last query, 1 minute by default on Standard and Jumbo, and bills at least a minute each time it starts
BackupsA restore point each day the data changed, kept 1 to 7 daysPoint-in-time restore with backups up to 90 days (Business)
UsersNo per-seat charge3 active users on Lite, 10 included on Business
Connectors15 sources synced on a schedule, at no extra chargeNot compared here

Which to choose

Which one fits.

When MotherDuck fits better

  • Your data is under 10 GB and your queries are light. Lite costs nothing, and DuckHouse has no free tier.
  • You want to pay query by query. Pulse charges for the compute each query uses, rather than for the time an instance is up.
  • You need more than 16 GB of memory, or read replicas. MotherDuck sells larger sizes, up to Giga, and read scaling with up to 16 replicas.
  • You need long backups. Business keeps point-in-time restores for up to 90 days; DuckHouse keeps restore points for 1 to 7.

When DuckHouse fits better

  • You’d rather not pay a plan fee. DuckHouse has none, so the first dollar goes to compute.
  • You want to know what you’re renting. Every DuckHouse size lists its vCPUs and memory, from 512 MB at $0.0117 an hour to 16 GB at $1.02.
  • You want your sources synced for you. DuckHouse keeps 15 of them in step, including Postgres, MySQL, Stripe, Shopify, Square, QuickBooks, Google Analytics, Meta Ads, at no extra charge.
  • You want a plain DuckDB server. Attach it from any DuckDB client over Quack, DuckDB’s own protocol, and keep your tables in a DuckDB file or as Parquet with DuckLake.

FAQ

About this comparison.

Where do these prices come from?
MotherDuck’s published list prices for its Tier 1 regions (AWS us-east-1 and us-west-2), checked on November 1, 2026 and linked at the foot of this page. DuckHouse’s figures are worked out by the same code that produces its invoices. Discounts, commitments and taxes are left out on both sides, and the examples count only the hours they state, leaving out per-start minimums and time spent idle before stopping.
Does DuckHouse have a free tier?
No. There is no free tier and no trial credit: you add a card before creating your first database, and pay for what runs. There is no subscription, minimum or contract either.
Why is DuckHouse priced at its largest size?
MotherDuck doesn’t say how many vCPUs or how much memory a Standard instance has, so there is no like-for-like DuckHouse size to put beside it. Using DuckHouse’s largest, 8 dedicated vCPUs and 16 GB, keeps the comparison from flattering DuckHouse.

Sources

Where the numbers come from.

MotherDuck’s prices are its published list prices for its Tier 1 regions (AWS us-east-1 and us-west-2), checked on November 1, 2026, and may have changed since. MotherDuck is a trademark of its owner. DuckHouse isn’t affiliated with or endorsed by MotherDuck.

Try it on your own data.

No contracts and no minimums. Pick a size, connect your sources, and stop the database whenever you like.