ScriptsMar 31, 2026·2 min read

Instructor — Structured Outputs from LLMs

Get structured, validated outputs from LLMs using Pydantic models. Works with OpenAI, Anthropic, Google, Ollama, and more. Retry logic, streaming, partial responses. 12.6K+ stars.

TO
TokRepo精选 · Community
Quick Use

Use it first, then decide how deep to go

This block should tell both the user and the agent what to copy, install, and apply first.

pip install instructor
import instructor
from pydantic import BaseModel
from openai import OpenAI

client = instructor.from_openai(OpenAI())

class User(BaseModel):
    name: str
    age: int
    bio: str

user = client.chat.completions.create(
    model="gpt-4o",
    response_model=User,
    messages=[{"role": "user", "content": "Extract: Jason is 25 and loves hiking."}],
)
print(user)  # User(name='Jason', age=25, bio='Loves hiking')
Intro

Instructor makes it easy to get structured, validated data from LLMs. Define a Pydantic model, and Instructor handles prompting, parsing, validation, and retries automatically. Works with OpenAI, Anthropic, Google, Ollama, LiteLLM, and any OpenAI-compatible API. Supports streaming, partial responses, and complex nested schemas. 12,600+ GitHub stars, MIT licensed.

Best for: Developers who need reliable structured data extraction from LLMs — not free-text Works with: OpenAI, Anthropic, Google, Ollama, LiteLLM, Mistral, Cohere

Key Features

Type-Safe Extraction

Define output schema with Pydantic, get validated objects back:

class Contact(BaseModel):
    name: str
    email: str
    company: Optional[str] = None

Automatic Retries

If the LLM returns invalid data, Instructor retries with the validation error as context:

client.chat.completions.create(
    response_model=Contact,
    max_retries=3,  # auto-retry on validation failure
    ...
)

Streaming & Partial Responses

Stream structured objects as they're generated — great for UIs.

Multi-Provider

One API across OpenAI, Anthropic, Google Gemini, Ollama, Mistral, and more.

Complex Schemas

Nested models, lists, enums, optional fields, custom validators — full Pydantic support.

FAQ

Q: What is Instructor? A: A Python library that gets structured, validated outputs from LLMs using Pydantic models. Handles prompting, parsing, validation, and retries. 12.6K+ stars.

Q: Does it work with Claude? A: Yes, Instructor supports Anthropic Claude, OpenAI, Google Gemini, Ollama, and many more providers.

🙏

Source & Thanks

Created by Jason Liu. Licensed under MIT. instructor-ai/instructor — 12,600+ GitHub stars

Related Assets

LobeChat — Modern AI Chat Framework & Agent Hub

Open-source AI chat framework with multi-agent collaboration, plugin marketplace, TTS, vision, and file upload. Supports 70+ model providers. Self-hostable. 75K+ stars.

TokRepo精选

AutoGPT — Autonomous AI Agent Platform

Build and deploy autonomous AI agents that accomplish goals with minimal human input. Visual builder, marketplace, and API. The original autonomous agent. 183K+ stars.

TokRepo精选

AnythingLLM — All-in-One AI Knowledge Base

All-in-one AI app: chat with documents, RAG, agents, multi-user, and 30+ LLM/embedding providers. Desktop + Docker. Privacy-first, no setup needed. 57K+ stars.

TokRepo精选
◈Home🏆Trending👤Me