xAI is an artificial intelligence company that develops large language models (LLMs). Their flagship model, Grok, is trained on real-time X (formerly Twitter) data and aims to provide witty, personality-rich responses while maintaining high capability on technical tasks. This guide will help you getting started with ChatXAI chat models. For detailed documentation of all ChatXAI features and configurations head to the API reference.

​
Overview

​
Integration details

ClassPackageLocalSerializablePY supportDownloadsVersion
ChatXAI@langchain/xai❌✅❌NPM - DownloadsNPM - Version

​
Model features

See the links in the table headers below for guides on how to use specific features.
Tool callingStructured outputJSON modeImage inputAudio inputVideo inputToken-level streamingToken usageLogprobs
✅✅✅❌❌❌✅✅✅

​
Setup

To access ChatXAI models you’ll need to create an xAI account, get an API key, and install the @langchain/xai integration package.

​
Credentials

Head to the xAI website to sign up to xAI and generate an API key. Once you’ve done this set the XAI_API_KEY environment variable: 
export XAI_API_KEY="your-api-key"
If you want to get automated tracing of your model calls you can also set your LangSmith API key by uncommenting below: 
# export LANGSMITH_TRACING="true"
# export LANGSMITH_API_KEY="your-api-key"

​
Installation

The LangChain ChatXAI integration lives in the @langchain/xai package: 

import IntegrationInstallTooltip from "@mdx_components/integration_install_tooltip.mdx";
<IntegrationInstallTooltip></IntegrationInstallTooltip>

<Npm2Yarn>
  @langchain/xai @langchain/core
</Npm2Yarn>

​
Instantiation

Now we can instantiate our model object and generate chat completions: 
import { ChatXAI } from "@langchain/xai"

const llm = new ChatXAI({
    model: "grok-beta", // default
    temperature: 0,
    maxTokens: undefined,
    maxRetries: 2,
    // other params...
})

​
Invocation

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

AIMessage {
  "id": "71d7e3d8-30dd-472c-8038-b6b283dcee63",
  "content": "J'adore programmer.",
  "additional_kwargs": {},
  "response_metadata": {
    "tokenUsage": {
      "promptTokens": 30,
      "completionTokens": 6,
      "totalTokens": 36
    },
    "finish_reason": "stop",
    "usage": {
      "prompt_tokens": 30,
      "completion_tokens": 6,
      "total_tokens": 36
    },
    "system_fingerprint": "fp_3e3898d4ce"
  },
  "tool_calls": [],
  "invalid_tool_calls": [],
  "usage_metadata": {
    "output_tokens": 6,
    "input_tokens": 30,
    "total_tokens": 36,
    "input_token_details": {},
    "output_token_details": {}
  }
}

console.log(aiMsg.content)

J'adore programmer.

​
Chaining

We can chain our model with a prompt template like so: 
import { ChatPromptTemplate } from "@langchain/core/prompts"

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

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

AIMessage {
  "id": "b2738008-8247-40e1-81dc-d9bf437a1a0c",
  "content": "Ich liebe das Programmieren.",
  "additional_kwargs": {},
  "response_metadata": {
    "tokenUsage": {
      "promptTokens": 25,
      "completionTokens": 7,
      "totalTokens": 32
    },
    "finish_reason": "stop",
    "usage": {
      "prompt_tokens": 25,
      "completion_tokens": 7,
      "total_tokens": 32
    },
    "system_fingerprint": "fp_3e3898d4ce"
  },
  "tool_calls": [],
  "invalid_tool_calls": [],
  "usage_metadata": {
    "output_tokens": 7,
    "input_tokens": 25,
    "total_tokens": 32,
    "input_token_details": {},
    "output_token_details": {}
  }
}
Behind the scenes, xAI uses the OpenAI SDK and OpenAI compatible API.

​
API reference

For detailed documentation of all ChatXAI features and configurations head to the API reference.