Web APIs in Python

FastAPI uses type hints and Pydantic models to validate requests, serialize responses and generate OpenAPI documentation automatically. It is async-first and built on Starlette. Inside Django, Django REST framework is the established toolkit, with serializers, viewsets, a browsable API and pluggable authentication and permissions. Django Ninja brings the FastAPI style of type-hinted endpoints and Pydantic schemas to Django, and django-modern-rest is a newer typed, async-capable option.

For Flask, Flask-RESTX adds resource classes and Swagger documentation, flask-smorest builds on marshmallow schemas to validate input and generate OpenAPI specs, and APIFlask wraps Flask with request validation and generated docs. Pick the one whose schema library matches what your project already uses. For a new API with no existing codebase, FastAPI or Django Ninja usually need the least code, because request validation and documentation both come from the same type hints, while Django REST framework is worth its extra ceremony when you want its permissions, throttling and generic views.

Beyond REST, Strawberry builds GraphQL schemas from type-annotated Python classes, with strawberry-django for Django models. Connexion is spec-first: you write the OpenAPI document and map its operations to Python functions. grpcio is the gRPC runtime for typed RPC between services over HTTP/2, with Protocol Buffers defining the contract. GraphQL fits clients that need to shape their own queries across related data, while gRPC fits internal service-to-service calls where both sides share generated code and you want strict schemas and streaming.