LLM Models and Integrations

What LLMs are, the providers LangChain supports, how the integration layer works and examples for OpenAI, Anthropic and local Ollama models.

05Prompts and Prompt Templates 11LangChain with LLM Providers 03Setting Up LangChain (Environment and Installation)

4.1

What are LLMs?

Large Language Models (LLMs) are AI models trained on massive text data.
They can understand, generate and manipulate human language.
Examples: OpenAI (GPT), Anthropic (Claude), Google (Gemini), Meta (Llama), etc.
LangChain provides easy integration with multiple LLM providers through a unified interface.

4.2

Supported LLM Providers in LangChain

OpenAI (GPT-3.5, GPT-4, GPT-4o)
Anthropic (Claude)
Google (Gemini)
HuggingFace (open-source models like Llama)
Ollama (local LLMs)
Cohere
Azure OpenAI
and many more via community integrations

4.3

Integration Architecture

Your Application

LangChainLLM interface

OpenAIGPT models

AnthropicClaude models

GoogleGemini models

HuggingFaceOpen-source models

OllamaLocal models

ResponseText, tokens, etc.

4.4

OpenAI Integration Example

python

from langchain_openai import ChatOpenAI

# Create LLM instance
llm = ChatOpenAI(
    model="gpt-3.5-turbo",
    temperature=0.7,
    api_key="your_api_key"
)

# Simple invocation
response = llm.invoke("Explain LangChain in simple words.")
print(response.content)

4.5

Anthropic (Claude) Integration Example

python

from langchain_anthropic import ChatAnthropic

llm = ChatAnthropic(
    model="claude-3-sonnet-20240229",
    temperature=0.7,
    api_key="your_api_key"
)

response = llm.invoke("What is LangChain?")
print(response.content)

4.6

Using a Local LLM with Ollama

python

from langchain_community.llms import Ollama

llm = Ollama(
    model="llama3",
    base_url="http://localhost:11434"
)

response = llm.invoke("Tell me a joke.")
print(response)

4.7

Key Parameters

Parameter	Description
model	Specify the LLM model name
temperature	Controls randomness (0.0 - 1.0)
api_key	API key for the provider
base_url	Required for local models (Ollama)
max_tokens	Maximum number of tokens in the response
top_p	Nucleus sampling parameter
stop	List of stop sequences