Python library · Mellow AI

Openextract

Structured data from documents and media, validated against your schema.

What I built

I built a library that turns documents, images, audio, and video into validated Pydantic objects, across model providers.

Designed a single extraction interface around Pydantic schemas, so callers receive validated objects rather than unstructured model responses.
Added synchronous and asynchronous APIs, multiple model providers, and inputs from files, URLs, or bytes.
Published a Python package with provider-specific installation options and explicit errors for provider failures and schema validation.

Extraction interface

from openextract import extract

invoice = extract(
    Invoice, "openai:gpt-5",
    source="invoice.pdf",
)