Skip to content

Marvin Documentation

Marvin is a Python library that lets you use Large Language Models by writing code, not prompts. It's open source, free to use, rigorously type-hinted, used by thousands of engineers, and built by the engineering team at Prefect.

Marvin is lightweight and is built for incremental adoption. You can use it purely as a serialization library and bring your own stack, or fully use its engine to work with OpenAI and other providers.

How Marvin feels

Marvin exposes a number of high level components to simplify working with AI. 

import marvin
from pydantic import BaseModel

class Location(BaseModel):
    city: str
    state: str
    latitude: float
    longitude: float

marvin.model(Location)("They say they're from the Windy City!")
# Location(city='Chicago', state='Illinois', latitude=41.8781, longitude=-87.6298)
Notice there's no code written, just the expected types. Marvin's components turn your function into a prompt, uses AI to get its most likely output, and parses its response.

Marvin exposes a number of high level components to simplify working with AI. 

from marvin import classifier
from typing import Literal

@classifier
def customer_intent(text: str) -> Literal['Store Hours', 'Pharmacy', 'Returns']:
    """Classifies incoming customer intent"""

customer_intent("I need to pick up my prescription") # "Pharmacy"
Notice customer_intent has no code. Marvin's components turn your function into a prompt, ask AI for its most likely output, and parses its response.

Marvin exposes a number of high level components to simplify working with AI. 

import marvin

@marvin.fn
def list_fruits(n: int, color: str = 'red') -> list[str]:
    """Generates a list of {{ n }} {{ color }} fruits"""

list_fruits(3) # "['Apple', 'Cherry', 'Strawberry']"
Notice list_fruits has no code. Marvin's components turn your function into a prompt, ask AI for its most likely output, and parses its response.

Learning Marvin

If you know Python, you already know Marvin. There are no fancy abstractions, just a handful of low-level, customizable decorators to give your existing code superpowers and a number of utilities that make your life as an AI Engineer easier no matter what framework you use.

Sections Description
Configuration Details on setting up Marvin and configuring various aspects of its behavior
AI Components Documentation for Marvin's familiar, Pythonic interfaces to AI-powered functionality.
API Utilities Low level API for building prompts and calling LLMs
Examples Deeper dives into how to use Marvin