Install required packages#

# Install SQLAlchemy: for ORM
pip install SQLAlchemy
# Optional asyncio
pip install "sqlalchemy[asyncio]"


# Install asyncpg: for async operations
pip install asyncpg


# Install Pydantic: for validation
pip install pydantic
# For email validation
pip install "pydantic[email]"



# Install fastAPI
pip install fastapi uvicorn

Follow the code structure#

app/ ├── main.py ├── db.py ├── models.py ├── schemas.py └── crud.py

Run to init model (database)#

python db.py

async def init_models():
    async with engine.begin() as conn:
        # Run the synchronous `create_all` within the async context
        await conn.run_sync(Base.metadata.create_all)

import asyncio


# Run the async function
if __name__ == "__main__":
    asyncio.run(init_models())

SQLAlchemy 2.1 Updates (2026)#

SQLAlchemy 2.1 (beta released January 2026) brings significant improvements:

Key Changes#

Feature	What Changed
Improved typing	`Row` type returns individual column types without `Tuple` wrapper
`filter_by()` scope	Now searches across all FROM clause entities, not just the primary
CREATE VIEW	Native DDL support for database views
Server-side UUIDv7	Monotonic UUID generation in batched INSERT operations
Free-threading	Accommodations for Python 3.14 free-threaded builds
`psycopg` (v3)	Improved PostgreSQL dialect for the modern async driver

SQLModel: FastAPI-Native Alternative#

For new API-first projects, consider SQLModel — it unifies Pydantic and SQLAlchemy models into a single class:

from sqlmodel import SQLModel, Field

# One class serves as both the database model AND the Pydantic schema
class User(SQLModel, table=True):
    id: int | None = Field(default=None, primary_key=True)
    name: str = Field(index=True)
    email: str = Field(unique=True)
    age: int | None = None

When to use SQLModel vs SQLAlchemy:

Scenario	Recommendation
New FastAPI project, simple models	SQLModel — less boilerplate, unified models
Complex relationships, legacy DB	SQLAlchemy — full control, mature ecosystem
Need raw SQL or advanced ORM features	SQLAlchemy — more flexible

References#