​
Bedrock


<Warning>
**You are currently on a page documenting the use of Amazon Bedrock models as [text completion models](/oss/javascript/concepts/text_llms). Many popular models available on Bedrock are [chat completion models](/oss/javascript/langchain/models).**


You may be looking for [this page instead](/oss/javascript/integrations/chat/bedrock/).
</Warning>
Amazon Bedrock is a fully managed service that makes Foundation Models (FMs) from leading AI startups and Amazon available via an API. You can choose from a wide range of FMs to find the model that is best suited for your use case.
This will help you get started with Bedrock completion models (LLMs) using LangChain. For detailed documentation on Bedrock features and configuration options, please refer to the API reference.

​
Overview

​
Integration details

ClassPackageLocalSerializablePY supportDownloadsVersion
Bedrock@langchain/community❌✅✅NPM - DownloadsNPM - Version

​
Setup

To access Bedrock models you’ll need to create an AWS account, get an API key, and install the @langchain/community integration, along with a few peer dependencies.

​
Credentials

Head to aws.amazon.com to sign up to AWS Bedrock and generate an API key. Once you’ve done this set the environment variables: 
export BEDROCK_AWS_REGION="your-region-url"
export BEDROCK_AWS_ACCESS_KEY_ID="your-access-key-id"
export BEDROCK_AWS_SECRET_ACCESS_KEY="your-secret-access-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 Bedrock integration lives in the @langchain/community package: 
import IntegrationInstallTooltip from "@mdx_components/integration_install_tooltip.mdx";
<IntegrationInstallTooltip></IntegrationInstallTooltip>

<Npm2Yarn>
  @langchain/community @langchain/core
</Npm2Yarn>

And install the peer dependencies:

<Npm2Yarn>
  @aws-crypto/sha256-js @aws-sdk/credential-provider-node @smithy/protocol-http @smithy/signature-v4 @smithy/eventstream-codec @smithy/util-utf8 @aws-sdk/types
</Npm2Yarn>

You can also use Bedrock in web environments such as Edge functions or Cloudflare Workers by omitting the `@aws-sdk/credential-provider-node` dependency
and using the `web` entrypoint:

<Npm2Yarn>
  @aws-crypto/sha256-js @smithy/protocol-http @smithy/signature-v4 @smithy/eventstream-codec @smithy/util-utf8 @aws-sdk/types
</Npm2Yarn>

​
Instantiation

Now we can instantiate our model object and generate chat completions: 
// @lc-docs-hide-cell
// Deno requires these imports, and way of loading env vars.
// we don't want to expose in the docs.
// Below this cell we have a typescript markdown codeblock with
// the node code.
import "@aws-sdk/credential-provider-node";
import "@smithy/protocol-http";
import "@aws-crypto/sha256-js";
import "@smithy/protocol-http";
import "@smithy/signature-v4";
import "@smithy/eventstream-codec";
import "@smithy/util-utf8";
import "@aws-sdk/types";
import { Bedrock } from "@langchain/community/llms/bedrock"
import { getEnvironmentVariable } from "@langchain/core/utils/env";

const llm = new Bedrock({
  model: "anthropic.claude-v2",
  region: "us-east-1",
  // endpointUrl: "custom.amazonaws.com",
  credentials: {
    accessKeyId: getEnvironmentVariable("BEDROCK_AWS_ACCESS_KEY_ID"),
    secretAccessKey: getEnvironmentVariable("BEDROCK_AWS_SECRET_ACCESS_KEY"),
  },
  temperature: 0,
  maxTokens: undefined,
  maxRetries: 2,
  // other params...
})

import { Bedrock } from "@langchain/community/llms/bedrock"

const llm = new Bedrock({
  model: "anthropic.claude-v2",
  region: process.env.BEDROCK_AWS_REGION ?? "us-east-1",
  // endpointUrl: "custom.amazonaws.com",
  credentials: {
    accessKeyId: process.env.BEDROCK_AWS_ACCESS_KEY_ID,
    secretAccessKey: process.env.BEDROCK_AWS_SECRET_ACCESS_KEY,
  },
  temperature: 0,
  maxTokens: undefined,
  maxRetries: 2,
  // other params...
})

​
Invocation

Note that some models require specific prompting techniques. For example, Anthropic’s Claude-v2 model will throw an error if the prompt does not start with Human:. 
const inputText = "Human: Bedrock is an AI company that\nAssistant: "

const completion = await llm.invoke(inputText)
completion

" Here are a few key points about Bedrock AI:\n" +
  "\n" +
  "- Bedrock was founded in 2021 and is based in San Fran"... 116 more characters

​
Chaining

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

const prompt = PromptTemplate.fromTemplate("Human: How to say {input} in {output_language}:\nAssistant:")

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

' Here is how to say "I love programming" in German:\n' +
  "\n" +
  "Ich liebe das Programmieren."

​
API reference

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