Skip to main content
THE RYDANTIC FIELD GUIDE

One model.Every boundary.

Give your data a contract. Carry it from JSON to APIs, databases and typed AI agents.

Pydantic-style modeling, built for Rust. Define a type once, then parse, validate, serialize and persist it with the integrations you need.

Start building Explore RydanticAI
model.rsRUST
Validated at the boundary.Ready for your application.
ONE RUST TYPEJSONJSON SchemaHTTPSQLCap’n ProtoAI
FROM IDEA TO IMPLEMENTATION

What are you building?

Pick a path. Keep the same model.

01 / CORE MODELS

Make your data trustworthy.

Parse JSON, normalize fields and return useful validation errors.

Your first model 02 / RYDANTIC AI

Build agents with a contract.

Typed tools, structured streams, image and PDF inputs, and a model gateway.

Build an agent 03 / HTTP APIS

Meet requests with validation.

Use Axum extractors, typed responses and clear error boundaries.

Create an endpoint 04 / STORAGE & WIRE FORMATS

Take your model further.

Connect SQL databases, evolve tables and generate native binary bindings.

Explore integrations
LEARN BY DOING

Start with a working project.

Run it. Change it. Make it yours.

GUIDED PROJECTS

Follow the tutorials.

Build a data contract, a CRUD API and a typed agent. Then add structured streams.

Start learning RUNNABLE RECIPES

Find your next example.

Copy a working command for models, HTTP, databases, agents and binary protocols.

Browse examples
KNOW YOUR VERSION

Core guides and the default installation follow Rydantic 0.7.0 stable, including its breaking explicit-derive contract. Callback lists, per-call context and pipelines are available in the opt-in 0.8.0-beta.1 prerelease; they are not included in 0.7.0. RydanticAI is independently versioned at 0.6.0, and PDF inputs remain AI development features. See release notes for breaking changes and publication status before upgrading.

Release notes
LOOKING FOR SOMETHING SPECIFIC?JSON Schema Cap’n Proto Cargo features