Soda vs Great Expectations

Quick verdict

Soda has two layers. Soda Core is a data contract verification engine: in version 4 you describe a dataset's expected schema and checks as a YAML contract and verify it from the CLI or Python API, locally, in a pipeline or through Soda Cloud. The README cites more than 50 built-in checks, with one package per data source. Soda Cloud stores contracts and results and adds metrics monitoring with anomaly detection, record-level anomaly detection, a diagnostics warehouse that keeps failed records in your own warehouse, alerting and a no-code interface. Version 3, with the SodaCL check language, is legacy.

GX Core is the current name of the Great Expectations Python library. You connect a Data Source (SQL database, Spark or Pandas dataframe, or files in cloud storage), define Batches, attach an Expectation Suite of declarative assertions, pair data and suite in a Validation Definition, and run it with a Checkpoint whose Actions can notify Slack, Microsoft Teams or email. Results render as Data Docs, HTML reports for people who do not read code. The documentation lists Python 3.10 to 3.14, states that Windows is not officially supported and names Airflow 2.9 or later as the only officially supported orchestrator.

Side by side

Key differences

Licences after the 2026 changes

For many teams this is now the first filter. GX Core remains Apache 2.0: you can use, modify, embed and redistribute it, including inside a service you sell. Soda Core is under the Elastic License 2.0. Soda's announcement says it stays source-available and free to use, and permits internal use, production use, modification, forking and professional services, but ELv2 forbids offering the software to third parties as a hosted or managed service, circumventing licence-key features, or removing licence notices. PyPI now classifies soda-core as proprietary. For an internal data team the practical effect is small; for a vendor building a data product, or a company whose policy requires OSI licences, it can rule Soda Core out.

Contracts in YAML versus Expectations in Python

Soda version 4 is built around data contracts: a YAML file that states the schema and the checks a dataset must satisfy, which a producing team and its consumers can agree on and keep in Git. That suits teams that want quality rules readable by people who do not write Python. GX Core expresses rules as Expectation classes in Python, which gives programmatic control (loops, parameters, custom Expectations) at the cost of more code. A minimal GX Core suite looks like this:

# Python (GX Core 1.x): two Expectations on an orders table
import great_expectations as gx

suite = gx.ExpectationSuite(name="orders_suite")
suite.add_expectation(gx.expectations.ExpectColumnValuesToNotBeNull(column="order_id"))
suite.add_expectation(gx.expectations.ExpectColumnValuesToBeBetween(column="amount", min_value=0))

Both approaches run against SQL databases by pushing queries to the engine. The equivalent idea in plain SQL is a query that returns the rows that break a rule, for example SELECT * FROM orders WHERE order_id IS NULL OR amount < 0; both tools generate and evaluate checks of this kind for you.

Hosted platform or run it yourself

With GX Cloud withdrawn, GX Core is a library you operate: you schedule Checkpoints in Airflow or another orchestrator, host Data Docs somewhere your users can reach, and wire Actions to your alert channels. There is no automatic anomaly detection or lineage in GX Core. Soda keeps a hosted path: Soda Cloud adds metrics monitoring and anomaly detection, alerting and ticketing, lineage and impact views, a no-code interface for business users on Team, and a Soda Runner (formerly Soda Agent) to run verifications inside your network. If you want monitoring and collaboration without building them, Soda has more to offer; if you want no vendor dependency at all, GX Core fits better.

Project health and migration risk

Both projects carry change risk. Soda users on version 3 face a move from SodaCL checks to version 4 contracts, which the repository describes as legacy versus current. GX Core users face a change of steward: Fivetran says the project will continue as open source and community driven, and releases continue (1.24.1 on PyPI, 22 October 2026), but long-term direction now depends on Fivetran and the community. The 0.18 line of Great Expectations reached end of life on 1 October 2025, so older code written for 0.x needs migration to the 1.x API either way.

Pricing and licensing

Soda. soda.io/pricing lists Free at $0 per month (unlimited users, pipeline testing, metrics observability, alerting and ticketing integrations, catalog integrations, and a free allowance of Soda Processing Units with extra SPUs pay as you go), Team at $750 per month (adds collaborative data contracts, a no-code interface, AI features, audit logs, custom roles and RBAC, private deployment, SSO and premium support; annual billing and volume discounts available) and Enterprise at custom pricing. The page uses a $ sign without naming the currency and does not state the size of the free SPU allowance. Soda Core itself is free to use under ELv2.

GX Core. Free under Apache 2.0; your costs are the infrastructure and the queries it runs. There is no hosted product to buy: Great Expectations announced that GX Cloud would no longer be publicly available from 1 June 2026. The greatexpectations.io pricing page still described GX Cloud plans on 7 November 2026, which conflicts with that announcement, so we do not list them as an option.

Pricing checked on the vendors' official pages on 7 November 2026. Prices change; confirm before buying.

Where each one leads

Limitations

When to choose each

When neither is right

• Your transformations are in dbt and you want tests next to the models: see Datafold vs dbt tests or Elementary.
• You validate dataframes only: Pandera (MIT) for pandas-style schemas or Deequ (Apache 2.0) for Spark are lighter options.
• You want automated monitoring with no rules to write: see Monte Carlo vs Bigeye.
• You run on AWS Glue or Databricks: AWS Glue Data Quality and Databricks data quality monitoring are built in.

Final recommendation

Frequently asked questions

Sources

• Soda pricing
• Soda Core licence update: moving to Elastic License 2.0
• Soda Core GitHub repository
• soda-core on PyPI
• Great Expectations: An Update for the Great Expectations Community (6 May 2026)
• GX Core GitHub repository
• great-expectations on PyPI
• GX docs: application integration support
• GX docs: GX Core overview

Checked November 2026.

How we research comparisons: our editorial method.