Environment:Anthropics Anthropic sdk python AWS Bedrock Environment

Knowledge Sources	anthropic-sdk-python AWS Bedrock
Domains	Infrastructure, Cloud_Provider
Last Updated	2026-02-15 12:00 GMT

Overview

AWS Bedrock provider environment requiring boto3/botocore and AWS credentials for accessing Claude models via Amazon Bedrock.

Description

This environment extends the core SDK environment with AWS-specific dependencies. It requires boto3 for AWS SDK access and botocore for low-level AWS service interaction. The Bedrock client infers the AWS region from environment variables or the boto3 session, defaulting to us-east-1 if unspecified. Authentication uses standard AWS credential chains (IAM roles, environment variables, credential files).

Usage

Use this environment when deploying Claude models through Amazon Bedrock instead of the direct Anthropic API. Required for the Cloud Provider Deployment workflow with Bedrock as the target provider.

System Requirements

Category	Requirement	Notes
OS	Linux, macOS, Windows	Same as core SDK
Python	>= 3.9	Same as core SDK
Network	HTTPS outbound to AWS Bedrock endpoints	Region-specific endpoints
AWS Account	Bedrock model access enabled	Must request model access in AWS console

Dependencies

Python Packages (Required)

All core SDK dependencies (see Environment:Anthropics_Anthropic_sdk_python_Python_SDK_Core_Environment)
boto3 >= 1.28.57
botocore >= 1.31.57

Credentials

The following environment variables are used:

AWS_REGION: AWS region for Bedrock API calls. Falls back to boto3 session region, then us-east-1.
AWS_ACCESS_KEY_ID: AWS access key (standard AWS credential chain).
AWS_SECRET_ACCESS_KEY: AWS secret key (standard AWS credential chain).
AWS_SESSION_TOKEN: Optional session token for temporary credentials.
ANTHROPIC_BEDROCK_BASE_URL: Override the Bedrock endpoint URL.

Quick Install

pip install 'anthropic[bedrock]'

Code Evidence

Optional dependency declaration from pyproject.toml:44:

bedrock = ["boto3 >= 1.28.57", "botocore >= 1.31.57"]

Region inference logic from bedrock/_client.py:70-90:

def _infer_region() -> str:
    aws_region = os.environ.get("AWS_REGION")
    if aws_region is None:
        try:
            import boto3
            session = boto3.Session()
            if session.region_name:
                aws_region = session.region_name
        except ImportError:
            pass

    if aws_region is None:
        log.warning("No AWS region specified, defaulting to us-east-1")
        aws_region = "us-east-1"  # fall back to legacy behavior

    return aws_region

API limitation checks from bedrock/_client.py:61-65:

if options.url.startswith("/v1/messages/batches"):
    raise AnthropicError("The Batch API is not supported in Bedrock yet")

if options.url == "/v1/messages/count_tokens":
    raise AnthropicError("Token counting is not supported in Bedrock yet")

Common Errors

Error Message	Cause	Solution
`MissingDependencyError: ... pip install anthropic[bedrock]`	boto3/botocore not installed	Run `pip install 'anthropic[bedrock]'`
`No AWS region specified, defaulting to us-east-1` (warning)	AWS_REGION not set and boto3 session has no region	Set `AWS_REGION` environment variable
`AnthropicError: The Batch API is not supported in Bedrock yet`	Batch API called via Bedrock client	Use direct Anthropic API for batch operations
`AnthropicError: Token counting is not supported in Bedrock yet`	count_tokens called via Bedrock	Use direct Anthropic API for token counting

Compatibility Notes

Batch API: Not supported through Bedrock. Use the direct Anthropic API client instead.
Token Counting: Not supported through Bedrock.
Region Fallback: If no region is specified, defaults to us-east-1 with a warning logged.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment