Skip to main content
Amazon Bedrock gives you access to foundation models from Anthropic, Meta, Mistral, and others through your AWS account. All data stays within your AWS environment.

Setup

Use your AWS credentials:
export AWS_ACCESS_KEY_ID=AKIA...
export AWS_SECRET_ACCESS_KEY=...
export AWS_REGION=us-east-1

Config

{
  "providers": {
    "amazon-bedrock": {
      "apiKey": "${AWS_ACCESS_KEY_ID}"
    }
  }
}

Use it

{
  "agents": [
    { "name": "coder", "model": "amazon-bedrock:us.anthropic.claude-sonnet-4-6-v2:0" }
  ]
}

Models

Bedrock model IDs use a different format with a region prefix:
ModelID on Bedrock
Claude Sonnetus.anthropic.claude-sonnet-4-6-v2:0
Claude Haikuus.anthropic.claude-haiku-4-5-20251001-v1:0
Llama 3.1 70Bus.meta.llama3-1-70b-instruct-v1:0
Mistral Largemistral.mistral-large-2407-v1:0
The us. and eu. prefixes in model IDs auto-infer to the amazon-bedrock provider. You can write us.anthropic.claude-sonnet-4-6-v2:0 without the provider prefix.

Features

FeatureSupported
StreamingYes
Tool useYes (model-dependent)
Vision (images)Yes (Claude models)
VPCYes
IAMYes
Data residencyYes

Provider Details

Provider IDamazon-bedrock
Env variablesAWS_ACCESS_KEY_ID + AWS_SECRET_ACCESS_KEY + AWS_REGION
API typeBedrock API
Auto-infer prefixesamazon., us., eu.

When to Use Bedrock

  • You’re already running infrastructure on AWS.
  • You need data to stay within your AWS VPC.
  • You want consolidated billing through AWS.
  • Your compliance requirements mandate AWS-specific controls.

Notes

  • Model availability varies by AWS region. Check the Bedrock model access page for your region.
  • You may need to request access to specific models in the Bedrock console before using them.
  • IAM permissions required: bedrock:InvokeModel and bedrock:InvokeModelWithResponseStream.