Pydantic | The end-to-end AI engineering stack

Trace every agent run, prove each fix with evals, and run the code your agents write in Agent Sandboxes.
Paste this into your coding agent
20 of the 25 largest companies on NASDAQ already rely on Pydantic in production.
Build with the open-source libraries. Deploy wherever you like. Logfire sits around the running system: watching every call, scoring every run, routing the models and running the code. Hosted by Pydantic, or self-hosted on Enterprise.
Audited controls, your data protected by design, and the documents your review needs.
New to Logfire: run the code your agents write in a secure, isolated interpreter, with no container to manage.
Python 3.15 support with lazy imports and TypeForm, a stable MISSING sentinel, frozendict support and faster schema generation.
Call TypeSafe’s Jev through AI Gateway with your own key, metered cost and input guardrails, on one key for every model.
Validate data, build agents, run their code, make HTTP calls. All MIT, all maintained by the same team.
Data validation from Python type hints. The one you already use.
Type-safe agent framework, with Pydantic Evals built in.
A sandboxed Python interpreter in Rust, for agent-written code.
Next-generation HTTP client. Sync and async, HTTP/1.1 and HTTP/2.
Evaluate agents with programmatic checks, LLM judges and human review.
Logfire SDKs, FastUI, Pydantic Settings, and the rest.
Our mission is simple: to accelerate engineers.
Sign up today and see your first trace in under 2 minutes
Free up to 10m records per month No credit card needed
© 2026 Pydantic Services Inc.