Pydantic | The end-to-end AI engineering stack

# Pydantic Pydantic is the AI engineering stack for teams building with Python. It covers the full cycle: validate data with the Pydantic library, build type-safe agents with Pydantic AI, route model calls with AI Gateway, and observe everything in production with Pydantic Logfire. Founded in 2018 by Samuel Colvin, backed by Sequoia, Partech, and Irregular. ## Products - **[Pydantic](https://github.com/pydantic/pydantic)** (OSS, MIT) — most widely used data validation library for Python; enforces type-hint-based validation at runtime; ~500M monthly downloads. Docs: https://docs.pydantic.dev - **[Pydantic AI](https://ai.pydantic.dev)** (OSS) — production-grade agent framework for Python; type-safe, model-agnostic, OpenTelemetry-native. Install: `pip install pydantic-ai` - **[Pydantic Logfire](https://logfire.pydantic.dev)** (SaaS) — AI observability and evals platform built on OpenTelemetry; ingests from any OTel-compatible language; first-party SDKs for Python (`pip install logfire`) and JavaScript/TypeScript (`npm install @pydantic/logfire-node`); features include traces, logs, metrics, LLM cost tracking, online evals, and an MCP server. - **[Pydantic Evals](https://ai.pydantic.dev/evals)** (OSS) — code-first evaluation library for LLM apps and agents; ships in `pydantic-ai` (Python) and the Logfire JS SDK; wire-compatible across both; integrates with Logfire for visualisation. - **[Pydantic AI Gateway](https://pydantic.dev/ai-gateway)** (SaaS) — unified LLM proxy; one key for all major providers, zero schema translation, granular spend caps, OTel audit trail. Consolidating into Logfire. ## Key URLs - [Homepage](https://pydantic.dev) - [Pydantic AI docs](https://ai.pydantic.dev) - [Logfire docs](https://logfire.pydantic.dev/docs) - [Pricing](https://pydantic.dev/pricing) - [Contact / demo](https://pydantic.dev/contact) - [GitHub org](https://github.com/pydantic) - [PyPI — pydantic](https://pypi.org/project/pydantic) - [PyPI — pydantic-ai](https:/