How to Beginner · 3 min read

How to use Pydantic AI with Anthropic

Q: How to use Pydantic AI with Anthropic

Use pydantic_ai with anthropic.Anthropic by creating an Anthropic client and passing it to instructor.from_anthropic(). Then define a pydantic.BaseModel for structured output and call client.chat.completions.create() with response_model to parse responses automatically.

Quick answer

Use pydantic_ai with anthropic.Anthropic by creating an Anthropic client and passing it to instructor.from_anthropic(). Then define a pydantic.BaseModel for structured output and call client.chat.completions.create() with response_model to parse responses automatically.

PREREQUISITES

Python 3.8+
Anthropic API key
pip install anthropic>=0.20 pydantic-ai>=0.1

Setup

Install the required packages and set your Anthropic API key as an environment variable.

Install packages: pip install anthropic pydantic-ai
Set environment variable: export ANTHROPIC_API_KEY='your_api_key' (Linux/macOS) or set ANTHROPIC_API_KEY=your_api_key (Windows)

bash

pip install anthropic pydantic-ai

output

Collecting anthropic
Collecting pydantic-ai
Successfully installed anthropic-0.20.0 pydantic-ai-0.1.0

Step by step

Use pydantic_ai with Anthropic by creating an Anthropic client and wrapping it with instructor.from_anthropic(). Define a pydantic.BaseModel to specify the expected structured response. Then call client.chat.completions.create() with response_model to get typed output.

python

import os
from anthropic import Anthropic
import instructor
from pydantic import BaseModel

# Initialize Anthropic client
client = Anthropic(api_key=os.environ["ANTHROPIC_API_KEY"])

# Wrap Anthropic client with Pydantic AI instructor
pydantic_client = instructor.from_anthropic(client)

# Define Pydantic model for structured extraction
class UserInfo(BaseModel):
    name: str
    age: int

# Prepare prompt
messages = [{"role": "user", "content": "Extract: John is 30 years old"}]

# Call Anthropic with response_model for structured output
response = pydantic_client.chat.completions.create(
    model="claude-3-5-sonnet-20241022",
    system="You are a helpful assistant.",
    messages=messages,
    response_model=UserInfo
)

# Access typed fields
print(f"Name: {response.name}, Age: {response.age}")

output

Name: John, Age: 30

Common variations

You can use async calls with pydantic_ai by awaiting the create() method. To use a different Anthropic model, change the model parameter accordingly. For OpenAI, use instructor.from_openai() similarly.

python

import asyncio

async def main():
    response = await pydantic_client.chat.completions.create(
        model="claude-3-5-sonnet-20241022",
        system="You are a helpful assistant.",
        messages=[{"role": "user", "content": "Extract: Alice is 25 years old"}],
        response_model=UserInfo
    )
    print(f"Name: {response.name}, Age: {response.age}")

asyncio.run(main())

output

Name: Alice, Age: 25

Troubleshooting

If you get a ValidationError, check that your BaseModel fields match the expected response format.
If the API returns unexpected text, ensure your prompt clearly instructs the model to respond in JSON matching your model.
Verify your ANTHROPIC_API_KEY environment variable is set correctly.

✅

Key Takeaways

Use instructor.from_anthropic() to integrate Pydantic AI with Anthropic easily.
Define a pydantic.BaseModel to parse structured responses automatically.
Always set response_model in chat.completions.create() for typed output.
Async calls are supported by awaiting the create() method.
Clear prompt instructions help avoid validation errors and parsing issues.

Verified 2026-04 · claude-3-5-sonnet-20241022

Verify ↗