For a complete example, see the Pydantic AI example on GitHub.
Pydantic AI uses OpenTelemetry internally, which Gentrace automatically captures when initialized, providing seamless tracing without additional configuration.

​
Prerequisites

​
Installation

pip install gentrace pydantic-ai pydantic-ai-openai

​
Configuration

Simply initialize Gentrace before using Pydantic AI:
pydantic_ai_simple.py
import os
from pydantic_ai import Agent
from pydantic_ai.models.openai import OpenAIModel
from gentrace import init, interaction

# Initialize Gentrace (will capture Pydantic AI's OTEL traces)
init(
    api_key=os.getenv("GENTRACE_API_KEY"),
    base_url=os.getenv("GENTRACE_BASE_URL", "https://gentrace.ai/api"),
)

# Create a simple Pydantic AI agent
agent = Agent(
    OpenAIModel("gpt-4o-mini"),
    system_prompt="You are a helpful assistant that gives concise answers.",
)

@interaction(name="pydantic_ai_chat", pipeline_id=os.getenv("GENTRACE_PIPELINE_ID", ""))
async def chat_with_agent(prompt: str) -> str:
    result = await agent.run(prompt)
    return result.output

# Usage
import asyncio

async def main():
    response = await chat_with_agent("What is 2+2?")
    print(f"Agent says: {response}")

asyncio.run(main())

​
Environment Variables

.env
GENTRACE_API_KEY=your-gentrace-api-key
GENTRACE_PIPELINE_ID=your-pipeline-id
OPENAI_API_KEY=your-openai-api-key

​
How It Works

Pydantic AI automatically generates OpenTelemetry traces for:
  • Agent invocations
  • Model calls
  • Tool usage
  • Retries and validation
Since Gentrace is OpenTelemetry-compatible, it automatically captures all these traces without requiring any additional instrumentation.

​
Supported Models

Pydantic AI supports multiple models through different providers:
  • OpenAI (GPT-4, GPT-3.5)
  • Anthropic (Claude)
  • Google (Gemini)
  • Groq
  • Mistral