ModelVault

Store Pydantic models directly

Schema-first typed storage for application models. Official CPython bindings for ModelVault (PyO3).

Store dataclasses and Pydantic v2 models with validation, indexes, migrations, and single-file deployment — no low-level schema JSON required for the recommended path.

Asyncio: AsyncDatabase runs engine work on a thread pool and supports concurrent reads on one handle (await asyncio.gather(...) for many get / query calls). Writes and transactions remain exclusive. See Async policy.

Read the docs: modelvault.readthedocs.io

Install

CPython 3.9+ · stable ABI wheels (cp39-abi3)

pip install "modelvault>=0.16.0,<0.17"

Quick start (Pydantic)

from pydantic import BaseModel
import modelvault

class Book(BaseModel):
    __modelvault_primary_key__ = "title"
    title: str
    year: int

db = modelvault.Database.open_in_memory()
books = modelvault.models.collection(db, Book)
books.insert(Book(title="ModelVault", year=2020))
print(books.get("ModelVault"))
print(modelvault.__version__)

Output:

title='ModelVault' year=2020
0.16.0

Dataclass example and indexed queries: Quickstart · Python guide

FastAPI: open with await AsyncDatabase.open(...), use modelvault.models.async_collection, and async def route handlers — FastAPI guide · runnable examples/fastapi_app.

Build from source

Requires Rust, Python 3.9+, and maturin.

cd python/modelvault && maturin develop --release && pytest -q

Contributor layout: python/README.md · full pipeline: make check-full from repo root

License

MIT — LICENSE