ChatGroq

This will help you getting started with Groq chat models. For detailed documentation of all ChatGroq features and configurations head to the API reference. For a list of all Groq models, visit this link.

Overview

Integration details

Class	Package	Local	Serializable	JS support	Package downloads	Package latest
ChatGroq	langchain-groq	❌	beta	✅

Model features

Tool calling	Structured output	JSON mode	Image input	Audio input	Video input	Token-level streaming	Native async	Token usage	Logprobs
✅	✅	✅	❌	❌	❌	✅	✅	✅	✅

Setup

To access Groq models you'll need to create a Groq account, get an API key, and install the langchain-groq integration package.

Credentials

Head to the Groq console to sign up to Groq and generate an API key. Once you've done this set the GROQ_API_KEY environment variable:

import getpass
import os

if "GROQ_API_KEY" not in os.environ:
    os.environ["GROQ_API_KEY"] = getpass.getpass("Enter your Groq API key: ")

If you want to get automated tracing of your model calls you can also set your LangSmith API key by uncommenting below:

# os.environ["LANGSMITH_API_KEY"] = getpass.getpass("Enter your LangSmith API key: ")
# os.environ["LANGSMITH_TRACING"] = "true"

Installation

The LangChain Groq integration lives in the langchain-groq package:

%pip install -qU langchain-groq

Instantiation

Now we can instantiate our model object and generate chat completions:

from langchain_groq import ChatGroq

llm = ChatGroq(
    model="llama-3.1-8b-instant",
    temperature=0,
    max_tokens=None,
    timeout=None,
    max_retries=2,
    # other params...
)

API Reference:ChatGroq

Invocation

messages = [
    (
        "system",
        "You are a helpful assistant that translates English to French. Translate the user sentence.",
    ),
    ("human", "I love programming."),
]
ai_msg = llm.invoke(messages)
ai_msg

AIMessage(content='The translation of "I love programming" to French is:\n\n"J\'adore le programmation."', additional_kwargs={}, response_metadata={'token_usage': {'completion_tokens': 22, 'prompt_tokens': 55, 'total_tokens': 77, 'completion_time': 0.029333333, 'prompt_time': 0.003502892, 'queue_time': 0.553054073, 'total_time': 0.032836225}, 'model_name': 'llama-3.1-8b-instant', 'system_fingerprint': 'fp_a491995411', 'finish_reason': 'stop', 'logprobs': None}, id='run-2b2da04a-993c-40ab-becc-201eab8b1a1b-0', usage_metadata={'input_tokens': 55, 'output_tokens': 22, 'total_tokens': 77})

print(ai_msg.content)

The translation of "I love programming" to French is:

"J'adore le programmation."

Chaining

We can chain our model with a prompt template like so:

from langchain_core.prompts import ChatPromptTemplate

prompt = ChatPromptTemplate.from_messages(
    [
        (
            "system",
            "You are a helpful assistant that translates {input_language} to {output_language}.",
        ),
        ("human", "{input}"),
    ]
)

chain = prompt | llm
chain.invoke(
    {
        "input_language": "English",
        "output_language": "German",
        "input": "I love programming.",
    }
)

API Reference:ChatPromptTemplate

AIMessage(content='Ich liebe Programmieren.', additional_kwargs={}, response_metadata={'token_usage': {'completion_tokens': 6, 'prompt_tokens': 50, 'total_tokens': 56, 'completion_time': 0.008, 'prompt_time': 0.003337935, 'queue_time': 0.20949214500000002, 'total_time': 0.011337935}, 'model_name': 'llama-3.1-8b-instant', 'system_fingerprint': 'fp_a491995411', 'finish_reason': 'stop', 'logprobs': None}, id='run-e33b48dc-5e55-466e-9ebd-7b48c81c3cbd-0', usage_metadata={'input_tokens': 50, 'output_tokens': 6, 'total_tokens': 56})

API reference

For detailed documentation of all ChatGroq features and configurations head to the API reference: https://python.langchain.com/api_reference/groq/chat_models/langchain_groq.chat_models.ChatGroq.html

Related

• Chat model conceptual guide
• Chat model how-to guides