Pydantic Models

Schemas: Zod / interfaces → Pydantic BaseModel

In TypeScript you describe a request body shape with an interface or a Zod schema. In FastAPI you describe it with a Pydantic BaseModel subclass. Pass that class as a type hint on your route function argument and FastAPI will parse and validate the request body automatically.

// TypeScript — Zod schema + inferred type
import { z } from 'zod';

const CreateItemSchema = z.object({
  name: z.string().min(1),
  price: z.number().positive(),
  inStock: z.boolean().default(true),
  tags: z.array(z.string()).default([]),
});

type CreateItem = z.infer<typeof CreateItemSchema>;

// NestJS DTO with class-validator
export class CreateItemDto {
  @IsString() @MinLength(1) name: string;
  @IsNumber() @IsPositive() price: number;
  @IsBoolean() @IsOptional() inStock?: boolean = true;
}

# FastAPI — Pydantic BaseModel
from pydantic import BaseModel
from typing import Optional, List

class CreateItem(BaseModel):
    name: str
    price: float
    in_stock: bool = True
    tags: List[str] = []

# Use it in a route — FastAPI parses + validates automatically
from fastapi import FastAPI
app = FastAPI()

@app.post("/items", status_code=201)
async def create_item(item: CreateItem):
    return item          # Pydantic serialises back to JSON

Nested models

Pydantic models compose naturally — embed one model inside another exactly like nested Zod objects.

// Zod nested object
const AddressSchema = z.object({
  street: z.string(),
  city: z.string(),
  zip: z.string().length(5),
});

const UserSchema = z.object({
  name: z.string(),
  address: AddressSchema,
});

# Pydantic nested models
from pydantic import BaseModel

class Address(BaseModel):
    street: str
    city: str
    zip_code: str

class User(BaseModel):
    name: str
    address: Address

# FastAPI will parse this nested JSON automatically:
# { "name": "Alice", "address": { "street": "1 Main St", ... } }

Response models — separating input from output

A common pattern is having separate models for input (what the client sends) and output (what you return). Pydantic makes this easy to compose.

// TypeScript — separate input / output types
interface CreateItemDto { name: string; price: number; }
interface ItemResponse { id: number; name: string; price: number; createdAt: string; }

# Pydantic — separate input / output models
from pydantic import BaseModel
from datetime import datetime

class CreateItem(BaseModel):       # Request body
    name: str
    price: float

class ItemResponse(BaseModel):     # Response shape
    id: int
    name: str
    price: float
    created_at: datetime

# Route uses response_model= to filter + serialize
from fastapi import FastAPI
app = FastAPI()

@app.post("/items", response_model=ItemResponse, status_code=201)
async def create_item(item: CreateItem):
    # Normally you'd save to DB and return the DB row
    return ItemResponse(id=1, name=item.name, price=item.price,
                        created_at=datetime.utcnow())

🟦 model_dump() and model_validate() — Pydantic v2 methods

Pydantic v2 (the current version) renamed the old .dict() → .model_dump() and .parse_obj() → .model_validate(). You may see legacy code with .dict() — it still works but is deprecated.

item = CreateItem(name="Widget", price=9.99)

# Serialize to dict
data = item.model_dump()
# {'name': 'Widget', 'price': 9.99, 'in_stock': True, 'tags': []}

# Deserialize from dict
item2 = CreateItem.model_validate({"name": "Gadget", "price": 4.99})

Try it — stdlib dataclasses as a runnable stand-in

Pydantic cannot run in the browser. dataclasses from the standard library gives you the same shape (typed fields, defaults, auto __init__) without validation, so you can explore the mental model here:

Open in an online REPL

from dataclasses import dataclass, field, asdict
from typing import Optional, List
import json

@dataclass
class CreateItem:
    name: str
    price: float
    in_stock: bool = True
    tags: List[str] = field(default_factory=list)
    description: Optional[str] = None

# Instantiate — same feel as BaseModel(...)
item = CreateItem(name="Widget", price=9.99, tags=["sale", "new"])
print("item:", item)

# Serialize to dict (like .model_dump())
data = asdict(item)
print("dict :", data)

# Serialize to JSON
print("json :", json.dumps(data, indent=2))

# Nested dataclass
@dataclass
class OrderLine:
    item: CreateItem
    quantity: int

line = OrderLine(item=item, quantity=3)
print("\nnested:", asdict(line))