Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Synthetic Data Vault

sdv.dev

  • Free plan
  • API
  • Self-hosted
7.0 Club score Visit sdv.dev

What Synthetic Data Vault is

Synthetic Data Vault (SDV) is a Python library and software platform for generating synthetic data from real data. It supports single-table, sequential, and relational multi-table data. Users can train synthesizers with different AI algorithms and generate data on demand. Quality tools help diagnose problems and compare synthetic results with real data. You can add constraints, change preprocessing, and choose anonymization options. SDV Community is a Python SDK for on-premises use, developed on Python 3.9–3.14. Its free plan includes five data types, five basic constraints, nine models, CSV and Excel support, local file types, and forum support, with limited commercial use. Add-on bundles include AI Connectors, which support metadata detection, data subsampling, and export in the original database format. Listed connectors include AlloyDB, BigQuery, MSSQL, Oracle, PostgreSQL, and Spanner. Differential Privacy uses an ε privacy-loss budget to manage privacy and quality tradeoffs. Listed paid options are bundles at $250 a month per bundle plus usage per user, and Enterprise Base at $500 a month plus usage per user.

Who it suits

SDV suits developers and data teams generating synthetic versions of single-table, sequential, or relational data. The Community SDK is for on-premises use; paid options add bundles or Enterprise features.

Good to know, and what to watch

  • Supports relational multi-table data
  • Includes tools to assess synthetic data quality
  • Community SDK can run on-premises
  • Differential Privacy offers a configurable privacy-loss budget
  • Community plan has limited commercial use
  • Bundles and Enterprise Base add usage charges
  • No separate paid-plan trial period is specified

SDV Community

Free

SDV Bundles

$250/mo

SDV Enterprise Base

$500/mo

Free plan

The verdict

SDV combines synthetic-data generation with quality evaluation and customization. Its free Community option has defined limits, while paid bundles and Enterprise Base add usage-based charges.

What the maker publishes

Customization
Users can add constraints, adjust data preprocessing, and select anonymization options for synthesizers. source
Database connectors
The AI Connectors page lists AlloyDB, BigQuery, MSSQL, Oracle, PostgreSQL, and Spanner. source
Data types
SDV supports single-table, sequential, and multi-table (relational) data. source
Deployment
SDV Community is a Python SDK that can be installed and used on-prem; the page says it was developed on Python 3.9–3.14. source
Evaluation
SDV includes tools to measure synthetic data quality, diagnose problems, and visualize comparisons with real data. source
Founded
2020 source
Integrations
The AI Connectors bundle supports database schema metadata detection, AI-ready data subsampling, and synthetic data export in the original database format. source
Modeling
Users can train synthesizers with different AI algorithms and generate synthetic data on demand. source
Privacy
The Differential Privacy bundle uses ε-differential privacy and lets users set a privacy-loss budget to control privacy and quality tradeoffs. source
Purpose
SDV is a Python library and software platform for creating synthetic data from real data. source
Support
The pricing page says Community and Enterprise users can get help through the DataCebo Forum, with Enterprise support prioritized during busy periods. source
Trial
The pricing FAQ says Community is free and can be used to evaluate SDV; it does not specify a separate paid-plan trial period. source
Usage billing
Enterprise usage-based pricing charges for synthetic data created and allows a monthly spending cap. source
Usage data
DataCebo's privacy policy says the software may collect and transmit usage data such as synthesizers used and the number of cells sampled for billing and service purposes. source

Checked against the maker's pages on 3 Oct 2026.