ORMs

Drizzle ORM33.5k★

Drizzle is the TypeScript ORM for developers who actually like SQL. Schema-as-code that looks like CREATE TABLE statements, zero code generation, instant type inference, and a 7.4KB bundle that makes serverless functions happy. It's what Prisma should have been if Prisma trusted developers to write queries. No .prisma schema files, no generated client, no waiting for prisma generate. Change your schema and types update immediately. The query builder gives you SQL control with full TypeScript safety. Drizzle Kit handles migrations, Drizzle Studio browses your data. Compared to Prisma (more abstraction, MongoDB support, bigger ecosystem), Drizzle is lighter and faster. Compared to Knex (query builder only, no type safety), Drizzle adds real TypeScript integration. Compared to raw SQL, Drizzle adds safety without losing control. Use this when you're building a TypeScript API and want SQL control with type safety, especially in serverless environments. Skip this if you need MongoDB or want maximum abstraction over database details. The catch: SQL-only — no MongoDB, no DynamoDB. The ecosystem is younger than Prisma's, so fewer tutorials and plugins. And if you don't know SQL, Drizzle's SQL-first approach will feel harder, not easier. Apache 2.0 license.

Sequelize30.4k★

Sequelize is the reliable old pickup truck of Node.js ORMs — it'll haul your data from A to B, but it's not pretty and it's not fast. It maps tables to JavaScript classes, supports every major SQL database, and has been doing it since before TypeScript was cool. Don't pick Sequelize for new TypeScript projects. Drizzle gives you SQL-close queries in 7KB with instant type inference. Prisma gives you the best DX with schema-first design and generated types. Sequelize v7 is still in alpha after years. The only reason to use it is maintaining a legacy codebase that already depends on it. The catch: Sequelize's TypeScript support is bolted on, not native. Migrations are fragile. The v6-to-v7 migration path is unclear, with v7 stuck in alpha since 2022. Community momentum has shifted to Drizzle and Prisma — Stack Overflow answers and tutorials increasingly ignore Sequelize. It works, but the ecosystem is moving on.

Kysely13.6k★

Kysely is a type-safe SQL query builder for TypeScript that actually trusts you to write SQL. No magic schema files, no ORM abstractions — just a query builder where TypeScript catches your column name typos at compile time. Drizzle ORM is the trendier alternative with built-in migrations and a schema DSL — most new projects default to it in 2026. Prisma is the ORM with the best DX for simple CRUD but generates bloated queries. Knex is the predecessor Kysely improves on, minus the type safety. Use Kysely if you're a SQL-first developer who wants full query control with TypeScript autocomplete. It shines when your queries are complex — joins, subqueries, CTEs — where ORMs generate garbage. Performance benchmarks show it slightly faster than Drizzle with less overhead. The catch: no built-in migrations or schema management. You'll pair it with a separate migration tool, which means more setup. And community momentum has shifted toward Drizzle — fewer tutorials, fewer integrations, smaller ecosystem. Kysely is the technically superior query builder that's losing the popularity contest.

SQLAlchemy11.7k★

SQLAlchemy is Python's database toolkit — not just an ORM, but a complete SQL abstraction layer that's been the foundation of serious Python backends for 20 years. It gives you two APIs: the ORM for when you want object-relational mapping, and Core for when you want to write SQL with Python's type safety. No other Python library offers that flexibility. If you're building anything database-heavy in Python — FastAPI services, data pipelines, analytics backends — SQLAlchemy is the default. Django ORM is simpler but welded to Django. Peewee is lightweight for small projects. Prisma has a Python client now but it's TypeScript-native. Tortoise ORM is the async alternative but with a fraction of the ecosystem. The catch: SQLAlchemy's power comes with complexity. The session management, identity map, and eager/lazy loading patterns have a real learning curve. Version 2.0 improved the API significantly, but legacy tutorials still pollute search results. And for simple CRUD, SQLAlchemy is overkill — you'll write more boilerplate than Django ORM for basic operations.