About Pydantic AI
Pydantic AI is a Python framework for building agents with typed inputs and outputs. It is designed so the same agent code can run in a web app, a terminal or a background job.
What it does
You define an agent with a model, instructions and tools, and the results are validated against the types you declare. Switching models is a small string change, and fallback models can take over when a provider fails. The wider project also covers evaluation, realtime voice and image generation, with separate packages for graphs and evals.
Key features
- Typed outputs validated against your types
- Model switching with a single string change
- Fallback models for provider failures
- Instrumentation based on OpenTelemetry
- Separate packages for evals and graphs
Who it's for
Python developers who want type checking and validation built into their AI code. It suits teams that already use Pydantic in their backend.
Pricing and good to know
The framework is installed as a Python package. Hosted products such as the Logfire platform and the AI Gateway are separate services with their own terms. Check those terms before sending production traffic through them.
Highlights
- Typed outputs validated against your types
- Model switching with a single string change
- Fallback models for provider failures
- Instrumentation based on OpenTelemetry
- Separate packages for evals and graphs
