See it in action
Pydantic AI: inspect a complete two-turn chat
The FastAPI example stores history between requests: a German Shepherd preference is followed by a recall question. This original screenshot includes both complete turns and replies, transcribed below with turn labels. The guide links the capture to its chat-app example without identifying a generation model.
Example input
User turn 1:
My favorite dog breed is the German Shepherd. What's yours?
User turn 2:
Can you remember what my favorite breed is?Example output
Assistant turn 1:
I don't have personal preferences or feelings, but I can help share information about various dog breeds if you'd like! German Shepherds are known for their intelligence, loyalty, and versatility. They're often used in working roles like police and service dogs. Is there anything specific you'd like to know about them or another breed?
Assistant turn 2:
You mentioned that your favorite dog breed is the German Shepherd.License
Complete original license and any source copyright retained. Source-authored examples are separate from site-reproduced results; no original scripts, installations, model/API requests, account creation, memory writes, graph queries or tests were executed. Complete visible messages/queries and selected replies are manually transcribed from the native author image; turn labels only identify sequence. Original PNG converted at native resolution to lossless WebP with identical decoded RGBA; no crop, retouch or upscaling.
The MIT License (MIT)
Copyright (c) Pydantic Services Inc. 2024 to present
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
Pydantic AI: preview the built-in web chat
The built-in browser UI brings conversation history, tool controls and model selection together. This original guide capture displays Claude Haiku 4.5 and the complete user question. The guide positions the UI for local development and debugging; production use calls for the documented authenticated frontend setup.
Example input
tell me something about yourselfLicense
Complete original license and any source copyright retained. Source-authored examples are separate from site-reproduced results; no original scripts, installations, model/API requests, account creation, memory writes, graph queries or tests were executed. Complete visible messages/queries and selected replies are manually transcribed from the native author image; turn labels only identify sequence. Original PNG converted at native resolution to lossless WebP with identical decoded RGBA; no crop, retouch or upscaling.
The MIT License (MIT)
Copyright (c) Pydantic Services Inc. 2024 to present
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
Pydantic AI: return a typed sentiment object
The author’s full script defines a typed Sentiment result, supplies a review-snippet tool, and asks about an app’s feedback. Its positive/0.9 source comment illustrates the output shape. Keep the declared provider/model and supply valid credentials when trying the recipe.
Example input
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=0.9Example output
label='positive' score=0.9License
Complete original license and any source copyright retained. Source-authored examples are separate from site-reproduced results; no original scripts, installations, model/API requests, account creation, memory writes, graph queries or tests were executed.
The MIT License (MIT)
Copyright (c) Pydantic Services Inc. 2024 to present
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
Product overview
Type-safe Python framework for building agents and LLM applications. Model-agnostic with built-in validation, structured outputs, and seamless observability.
What Pydantic AI offers
- 01Validate structured model outputs
- 02Connect agent functions and MCP tools
- 03Trace runs and evaluate agent behavior
What makes it different
Python type definitions are used to describe outputs and callable tools. Provider costs and observability services are separate from the open-source framework.
Who should consider it
- Python application developers
- Teams building agents with typed data
What to verify
- Try one real task and compare its output with your requirements.
- Check what data the service stores, shares, or uses for training.
- Confirm current pricing, usage limits, and cancellation terms on the official site.
Official link checked on Nov 9, 2026. Feature claims have not been independently tested.
