whichdevtool?
Menu
StacksDirectory Pricing changes Glossary
VPS and cloud servers App hosting and PaaS Managed databases Transactional email APIs Email marketing and newsletters Domain registrars CDN, DNS and edge security Monitoring, uptime and error tracking Auth and identity Frontend and full-stack frameworks Backend frameworks UI component libraries AI model APIs File and object storage Payments and billing Background jobs, queues and workflows ORMs and database toolkits

Python SQL toolkit and ORM · by SQLAlchemy (community project)

SQLAlchemy

Python's standard SQL toolkit and ORM: a Core expression language plus a unit-of-work ORM, with Alembic for migrations and dialects for most SQL databases; version 2.1 shipped in September 2026.

Visit sqlalchemy.org ↗ Alternatives Open source Open source Self-hostable
SQLAlchemy homepage
§1

Overview

SQLAlchemy has been developed by its authors and contributors since at least 2007 (the start of its copyright notice) and is the most widely used database library in Python. Core gives you a SQL expression language, schema metadata and connection pooling; the ORM adds declarative mapped classes, relationships, a Session with identity map and unit of work, and many loading strategies. Since 2.0, queries use a unified select() style with type-annotated Mapped[] columns that editors and type checkers understand.

Schema migrations are handled by Alembic, a separate tool from the same project, which can autogenerate migration scripts by comparing models with the database; its docs describe what autogenerate does and does not detect.

SQLAlchemy 2.1.0 was released on 24 September 2026, three years after 2.0. It requires Python 3.11, no longer installs greenlet by default (install sqlalchemy[asyncio] for async), defaults postgresql:// URLs to psycopg 3 and oracle:// to oracledb, and adds t-string SQL, CREATE VIEW support and Microsoft's mssql-python driver. The project says the upgrade is far smaller than 1.4 to 2.0.

§2

Pricing and plans

Plan Price What you get
SQLAlchemy $0 MIT licence

Free and open source; there is no paid product.

Source: www.sqlalchemy.org/ · prices as published, excluding tax.

§3

What it does

Core expression language
Build SQL with Python objects, independent of the ORM, with full control over the statement.
ORM with unit of work
Session tracks changes to mapped objects and flushes them in dependency order.
Typed mappings
Declarative classes with Mapped[] annotations are understood by mypy and IDEs.
Asyncio support
AsyncEngine and AsyncSession for asyncpg, psycopg and other async drivers (install the asyncio extra in 2.1).
Alembic migrations
Autogenerate revision scripts from model changes, branch and merge revisions, and generate offline SQL.
Many dialects
Built-in PostgreSQL, MySQL/MariaDB, SQLite, Oracle and SQL Server, plus external dialects for Snowflake, BigQuery, ClickHouse, CockroachDB and more.
§4

Strengths and limitations

Strengths

  • Mature and extremely well documented, with two decades of production use
  • Lets you drop to Core SQL whenever the ORM gets in the way
  • Wide dialect coverage through built-in and third-party dialects
  • Foundation for SQLModel, Flask-SQLAlchemy and many other tools

Limitations

  • Large API with two layers (Core and ORM) and many loading options, so the learning curve is steep
  • Migrations need a separate tool (Alembic), and autogenerate does not detect every change
  • Async support relies on greenlet, which 2.1 no longer installs by default
  • 2.1 requires Python 3.11 or newer
§5

Who it suits

Good for

  • Python web apps on FastAPI, Flask or Starlette
  • Data-heavy services that mix ORM and hand-written SQL
  • Teams that need Oracle, SQL Server or unusual databases from Python

Look elsewhere if

  • Django projects, which have their own ORM
  • Very small scripts where a lighter library such as Peewee is enough
§6

Pricing history

  • SQLAlchemy 2.1 released

    SQLAlchemy 2.1.0 requires Python 3.11, no longer installs greenlet by default, defaults to psycopg 3 and oracledb, and adds t-string SQL, CREATE VIEW and the mssql-python driver. Source ↗

§7

Alternatives to SQLAlchemy

Tool From Free option Stages
Active Record (Rails) Active Record ORM for Ruby on Rails
— Open source
Eloquent ORM Active Record ORM for Laravel
— Open source
Dapper .NET micro-ORM
— Open source
— Open source
Hibernate ORM Java ORM and Jakarta Persistence implementation
— Open source
Doctrine ORM Data mapper ORM for PHP
— Open source

All ORMs and database toolkits →

§8

Questions

How do I do migrations with SQLAlchemy?

Use Alembic, the migration tool from the same project; it can autogenerate revision scripts from your models.

What changed in SQLAlchemy 2.1?

Python 3.11 minimum, greenlet no longer installed by default, psycopg 3 and oracledb as default drivers, plus t-string SQL, CREATE VIEW and the mssql-python driver.

Does SQLAlchemy support async?

Yes, through AsyncEngine and AsyncSession; in 2.1 install sqlalchemy[asyncio] to get the greenlet dependency.

SQLAlchemy vs SQLModel: which should I use?

SQLModel is a thinner layer on SQLAlchemy and Pydantic for FastAPI apps; SQLAlchemy exposes the full feature set directly.

Sources read for this page