Cohere integration

Integrate with the Cohere LLM using LangChain Python.

You are currently on a page documenting the use of Cohere models as text completion models. Many popular Cohere models are chat completion models.

You may be looking for this page instead.

Cohere is a Canadian startup that provides natural language processing models that help companies improve human-machine interactions.

Head to the API reference for detailed documentation of all attributes and methods.

Overview

Integration details

Class	Package	Local	Serializable	JS support	Downloads	Version
Cohere	langchain-community	❌	beta	✅

Setup

The integration lives in the langchain-community package. We also need to install the cohere package itself. We can install these with:

Credentials

We’ll need to get a Cohere API key and set the COHERE_API_KEY environment variable:

import getpass
import os

if "COHERE_API_KEY" not in os.environ:
    os.environ["COHERE_API_KEY"] = getpass.getpass()

Installation

pip install -U langchain-community langchain-cohere

It’s also helpful (but not needed) to set up LangSmith for best-in-class observability

os.environ["LANGSMITH_TRACING"] = "true"
# os.environ["LANGSMITH_API_KEY"] = getpass.getpass()

Invocation

Cohere supports all LLM functionality:

from langchain_cohere import Cohere
from langchain.messages import HumanMessage

model = Cohere(max_tokens=256, temperature=0.75)

message = "Knock knock"
model.invoke(message)

" Who's there?"

await model.ainvoke(message)

" Who's there?"

for chunk in model.stream(message):
    print(chunk, end="", flush=True)

 Who's there?

model.batch([message])

[" Who's there?"]

---

API reference

For detailed documentation of all Cohere llm features and configurations head to the API reference: python.langchain.com/api_reference/community/llms/langchain_community.llms.cohere.Cohere.html

---

Edit this page on GitHub or file an issue.

Connect these docs to Claude, VSCode, and more via MCP for real-time answers.