Pydantic AI
A Python SDK for typed agent loops, validated structured responses and function-based tools.
About this listing
Pydantic AI helps Python developers build agent applications around typed inputs, tools and outputs. An agent can be given a Pydantic output model so that its result is validated into the shape the application expects. Python functions can become tools, with argument validation and access to dependencies through a run context.
The SDK supports multiple model providers and documents fallback handling and durable-execution integrations. Start by selecting a model, declaring the output type and registering the functions the agent may call, then run it from your application. This is a development toolkit, not a hosted finished agent. Free refers to the MIT-licensed code; model API calls, infrastructure and optional hosted services can carry separate charges.