Langchain-Cohere

This package contains the LangChain integrations for Cohere.

Cohere empowers every developer and enterprise to build amazing products and capture true business value with language AI.

Installation

• Install the langchain-cohere package:

pip install langchain-cohere

• Get a Cohere API key and set it as an environment variable (COHERE_API_KEY)

Migration from langchain-community

Cohere's integrations used to be part of the langchain-community package, but since version 0.0.30 the integration in langchain-community has been deprecated in favour langchain-cohere.

The two steps to migrate are:

1. Import from langchain_cohere instead of langchain_community, for example:

   • from langchain_community.chat_models import ChatCohere -> from langchain_cohere import ChatCohere
   • from langchain_community.retrievers import CohereRagRetriever -> from langchain_cohere import CohereRagRetriever
   • from langchain.embeddings import CohereEmbeddings -> from langchain_cohere import CohereEmbeddings
   • from langchain.retrievers.document_compressors import CohereRerank -> from langchain_cohere import CohereRerank

2. The Cohere Python SDK version is now managed by this package and only v5+ is supported.

   • There's no longer a need to specify cohere as a dependency in requirements.txt/pyproject.toml (etc.)

Supported LangChain Integrations

Usage Examples

Chat

from langchain_cohere import ChatCohere
from langchain_core.messages import HumanMessage

llm = ChatCohere()

messages = [HumanMessage(content="Hello, can you introduce yourself?")]
print(llm.invoke(messages))

ReAct Agent

from langchain_community.tools.tavily_search import TavilySearchResults
from langchain_cohere import ChatCohere, create_cohere_react_agent
from langchain.prompts import ChatPromptTemplate
from langchain.agents import AgentExecutor

llm = ChatCohere()

internet_search = TavilySearchResults(max_results=4)
internet_search.name = "internet_search"
internet_search.description = "Route a user query to the internet"

prompt = ChatPromptTemplate.from_template("{input}")

agent = create_cohere_react_agent(
    llm,
    [internet_search],
    prompt
)

agent_executor = AgentExecutor(agent=agent, tools=[internet_search], verbose=True)

agent_executor.invoke({
    "input": "In what year was the company that was founded as Sound of Music added to the S&P 500?",
})

RAG Retriever

from langchain_cohere import ChatCohere, CohereRagRetriever

rag = CohereRagRetriever(llm=ChatCohere())
print(rag.get_relevant_documents("Who are Cohere?"))

Text Embedding

from langchain_cohere import CohereEmbeddings

embeddings = CohereEmbeddings(model="embed-english-light-v3.0")
print(embeddings.embed_documents(["This is a test document."]))

Contributing

Contributions to this project are welcomed and appreciated. The LangChain contribution guide has instructions on how to setup a local environment and contribute pull requests.