OpenAI Integration
Instructor works seamlessly with OpenAI models. Here's how to use it with different OpenAI features.
import instructor
from openai import OpenAI
from pydantic import BaseModel
class User(BaseModel):
name: str
age: int
client = instructor.from_openai(OpenAI())GPT-3.5 Turbo (cheaper, faster)
user = client.chat.completions.create(
model="gpt-3.5-turbo",
response_model=User,
messages=[
{"role": "user", "content": "Extract: John is a 25-year-old engineer."}
]
)GPT-4 (more capable, better at complex tasks)
user = client.chat.completions.create(
model="gpt-4",
response_model=User,
messages=[
{"role": "user", "content": "Extract: John is a 25-year-old engineer."}
]
)
user = client.chat.completions.create(
model="gpt-3.5-turbo",
response_model=User,
messages=[
{"role": "system", "content": "You are an expert at data extraction."},
{"role": "user", "content": "Extract user details from: John is 25 years old."}
]
)Lower temperature for more consistent results
user = client.chat.completions.create(
model="gpt-3.5-turbo",
temperature=0.1, # Very deterministic (0.0-1.0)
response_model=User,
messages=[
{"role": "user", "content": "Extract: John is a 25-year-old engineer."}
]
)Instructor defaults to using OpenAI's function calling format, but you can use JSON mode too:
client = instructor.from_openai(
OpenAI(),
mode=instructor.Mode.JSON
)
user = client.chat.completions.create(
model="gpt-3.5-turbo",
response_model=User,
messages=[
{"role": "user", "content": "Extract: John is a 25-year-old engineer."}
]
)
import asyncio
from openai import AsyncOpenAI
async_client = instructor.from_openai(AsyncOpenAI())
async def extract_user():
return await async_client.chat.completions.create(
model="gpt-3.5-turbo",
response_model=User,
messages=[
{"role": "user", "content": "Extract: John is a 25-year-old engineer."}
]
)
user = asyncio.run(extract_user())Running the Example
First, install Instructor and any dependencies
$ pip install instructor pydantic
Run the Python script
$ python openai.py