The license is the smaller half of the bill. The bigger half is who writes the tests, and the two halves trade against each other: the cheapest software usually asks for the most typing.
What each option costs, with dates
Prices below are dated because they move. DataKitchen’s own prices are from the pricing page as listed in October 2026. Vendor ranges are DataKitchen’s estimates from the comparison pages, reviewed June 2026 and sized for 10 users and 1,000 tables over one year. Vendors that don’t publish rates are estimated from reported deal sizes, so treat those as ranges, not quotes.
| Option | Pricing model | What you pay | Source |
|---|---|---|---|
| Open-source DataOps TestGen | Apache 2.0 | $0, for 1 user, 1 connection, 1 project | Pricing page |
| Open-source DataOps Observability | Apache 2.0 | $0, for 1 user, 1 project | Pricing page |
| TestGen Enterprise | Per user and per database connection | $100 per user per month plus $100 per connection per month | Pricing page |
| DataOps Observability Enterprise | Per user and per agent | $100 per user per month plus $100 per agent per month | Pricing page |
| Recce | Open core | $0 open source; Team tier $250 a month | Comparison page |
| AWS Glue Data Quality | Per DPU-hour | $5,000 to $30,000 a year, usage-based (DataKitchen’s estimate) | Comparison page |
| Metaplane (Datadog) | Per table and consumption | $60,000 to $150,000 a year (DataKitchen’s estimate) | Comparison page |
| Bigeye | Per-table tier | $80,000 to $180,000 a year (DataKitchen’s estimate) | Comparison page |
| Soda Enterprise | Per dataset and user | $100,000 to $200,000 a year (DataKitchen’s estimate) | Comparison page |
| Monte Carlo | Per-table tier | $120,000 to $250,000 a year (DataKitchen’s estimate) | Comparison page |
A separate figure comes from an earlier DataKitchen post: across the enterprise vendors on the comparison pages, the average came to about $172,000 a year as of April 2026. That average predates the June 2026 review behind the ranges above, so the two won’t reconcile exactly.
A worked example for a team of four
Say you have four people and two databases, and you want profiling, generated tests, and table monitors on everything.
Open-source TestGen covers one of you on one database for $0. The second person or the second database is where the free edition stops, and that limit is real. TestGen Enterprise for the whole team is 4 users plus 2 connections, six units at $100 a month: $600 a month, or $7,200 a year. Tables don’t change the number. Ten users on three databases comes to $15,600 a year, which is the figure the pricing page uses.
The per-table version of the same team looks different. In Stop Paying The Data Quality Tax, Chris Bergh describes a customer with a six-person data team, 1,000 tables in the main application, and 500 more in the warehouse. Soda would have cost $750 a month base plus per-table fees adding up to nearly $12,000 a month, over $150,000 a year. DataKitchen costs that team $9,000.
Our explanation for the six-figure prices is funding. More than $1 billion in venture capital has gone into data observability and data quality vendors, and that capital has to return a multiple. Price is where it comes from.
The cost that “free” leaves on your desk
Open-source frameworks such as dbt tests, Soda Core, and Great Expectations (GX Core, now stewarded by Fivetran; GX Cloud shut down June 1, 2026) are $0 to download and expect you to write each rule yourself, in SQL, YAML, or Python, as our post on generating 80% of tests lays out. Count the hours before you call that free. The Guide to DataOps Tests works the math: hand-writing 2,500 tests at 30 minutes each takes 1,250 hours. That is more than half a working year for one engineer, and the tests still need upkeep when the data changes.
So the honest comparison is license plus labor. A tool that generates most tests from a profile moves the cost from your calendar to a line item. A tool that charges per table moves it back, because you start choosing which tables to leave unwatched.
A benchmark you can take into a budget meeting
DataKitchen’s rule, stated on the pricing page: a year of data quality tooling should cost about one data engineer’s monthly salary, not their annual salary. For a team of four on a few databases, the February 2026 post puts that at $0, or at most $9,000 to $10,000 a year.
We sell one of the products in that table, so check our math. Run an open-source tool on your own data first. Then you’re comparing a quote against a working baseline instead of against a demo of an empty environment.