DuckDB

Open source

An analytical, in-process SQL database that runs inside your application or from a CLI, with a rich SQL dialect and easy CSV and Parquet querying.

duckdb.org
DuckDB homepage screenshot
GitHub stars
43k
Last 30 days
+900(+2%)
Last commit
today
Repository age
8 years
Version
v2.0.0
Licence
MIT
Self-hosted
Yes

About DuckDB

DuckDB is an analytical SQL database management system that runs in-process, meaning it is embedded in your application or notebook rather than running as a separate server. The goals are speed, reliability, portability and ease of use for analytical queries, and it is written in C++.

Its SQL dialect goes well beyond the basics: it handles arbitrary nested correlated subqueries, window functions, collations and complex types such as arrays, structs and maps, plus extensions that make SQL more convenient. CSV and Parquet files can be queried by naming them in a FROM clause. DuckDB is available as a standalone command line tool and has clients for Python, R, Java and WebAssembly, with deep integration into packages such as pandas and dplyr.

The project is MIT licensed. Building from source needs CMake, Python 3 and a C++17 compiler. Because it needs no server, DuckDB is popular with data analysts and engineers for exploring files and running local analytics, and it complements larger warehouses rather than replacing them.

Key features

  • In-process analytical SQL engine
  • Direct querying of CSV and Parquet files
  • Window functions and correlated subqueries
  • Arrays, structs and maps data types
  • CLI plus Python, R, Java and Wasm clients
  • Integrations with pandas and dplyr

Good fit for

  • Analyzing local files with SQL
  • Embedded analytics inside applications
  • Data science workflows in Python or R
Built with
C++ tools
Tags
database
olap
analytics
sql
embedded-database
parquet
python
in-process

DuckDB: questions and answers

What is DuckDB used for?
DuckDB is an analytical, in-process SQL database that runs inside your application or from a CLI, with a rich SQL dialect and easy CSV and Parquet querying. It is a good fit for analyzing local files with SQL, embedded analytics inside applications and data science workflows in Python or R.
Is DuckDB open source?
Yes. DuckDB is open source under the MIT licence. Its source code is on GitHub at duckdb/duckdb and is written mainly in C++.
Is DuckDB free?
Yes. DuckDB is open source, so the software itself is free to use.
Can I self-host DuckDB?
Yes. DuckDB can be self-hosted on your own server or infrastructure; there is no official hosted version.
What is DuckDB an alternative to?
DuckDB is an open-source alternative to MotherDuck, Amazon Athena, Tinybird and Microsoft SQL Server. Other open-source alternatives to MotherDuck include ClickHouse, TimescaleDB and StarRocks.
Is DuckDB actively maintained?
Yes. The most recent commit to DuckDB was on 12 November 2026, and the latest release is v2.0.0, published on 21 October 2026. The project has 43k stars on GitHub, with 900 added in the last 30 days.

Open-source alternatives to DuckDB

See all

SaaS alternatives to DuckDB

See all