pydantic-ai
How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed end to end.
What is pydantic-ai?
How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed end to end.
Overview of pydantic-ai
What it is
Pydantic AI is a Python AI SDK described in its README as "How Python does AI. Agents, realtime voice, image generation, embeddings. Every model, every interface, typed end to end." It provides a typed, extensible agent loop where every model is a string swap away and another can be used as a fallback when a provider fails. The same agent can run behind a web frontend, in the terminal, on a voice call, on a durable background queue, in GitHub Actions, or as a plain object with a run() method. The project is open source under the MIT license.
Key features
- Typed agent loop with structured output validation via Pydantic models.
- Model swapping and fallback: every model is a string swap away, with automatic fallback to another model on failure.
- Realtime voice sessions using OpenAI Realtime, Gemini Live, Azure, and xAI Grok Voice.
- Image generation with dedicated image models, either standalone or as part of an agent run.
- Embeddings support included in the same package.
- Durable execution via integrations with Temporal, DBOS, Prefect, Restate, AWS Lambda, Airflow, and others.
- Capabilities system (memory, guardrails, sub-agents, planning, context management, persistence) via the separate Pydantic AI Harness package.
Use cases
- Simple typed data extraction from text.
- Complex, long-running multi-agent collaboration.
- Coding agent with file access, shell, repo context, planning, and context management.
- Durable workflows that survive restarts and failures.
- Realtime voice assistants with tool calling.
- Image generation for applications that need to produce visuals.
Getting started
Install the package and run a basic data extraction agent:
uv add pydantic-ai
from typing import Literal
from pydantic import BaseModel, Field
from pydantic_ai import Agent, RunContext
class Sentiment(BaseModel):
label: Literal['positive', 'negative', 'neutral']
score: float = Field(ge=-1, le=1)
agent = Agent('openai:gpt-6.1-sol', output_type=Sentiment)
@agent.tool
def recent_reviews(ctx: RunContext, product: str) -> list[str]:
"""Fetch recent review snippets for a product."""
return ['The new release fixed everything I complained about!']
result = agent.run_sync('How are people feeling about the Extract app?')
print(result.output)
### > label='positive' score=0.9
License and pricing
MIT license, open source.
Based on the project's public README and repository metadata.