FastAPI

FastAPI is a modern Python web framework built for building APIs fast — both in development speed and runtime performance. It uses Python type hints to drive automatic validation, serialization, and OpenAPI documentation, eliminating an entire category of boilerplate. Under the hood it’s built on Starlette (ASGI) and Pydantic, making it fully async-capable and production-ready with Uvicorn or Gunicorn. It has become the go-to Python API framework for ML serving, microservices, and greenfield REST APIs.

Quick start

pip install fastapi uvicorn

# main.py
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel

app = FastAPI()

class Item(BaseModel):
    name: str
    price: float
    in_stock: bool = True

items: dict[int, Item] = {}

@app.get("/")
async def root():
    return {"message": "Hello from FastAPI"}

@app.get("/items/{item_id}")
async def get_item(item_id: int) -> Item:
    if item_id not in items:
        raise HTTPException(status_code=404, detail="Item not found")
    return items[item_id]

@app.post("/items", status_code=201)
async def create_item(item: Item) -> Item:
    item_id = len(items) + 1
    items[item_id] = item
    return item

# Run the server — docs at http://localhost:8000/docs
uvicorn main:app --reload

When to use

FastAPI is the best choice for Python REST APIs and ML model serving endpoints. Its combination of auto-docs, Pydantic validation, and async performance makes it ideal for data science teams exposing models or microservice backends. For a full-featured web application with an admin panel, Django is the better choice. For very simple scripts-turned-servers, Flask has less setup overhead.