AWS Bedrock Configuration

Overview

This guide explains how to enable AWS Bedrock foundation models in your AWS account for use with AI/Run CodeMie. AWS Bedrock provides access to high-performance foundation models from Anthropic (Claude), Amazon (Titan), etc.

When to Use This Guide

This configuration is required if you plan to use AWS Bedrock models such as Anthropic Claude, Amazon Titan, or other foundation models available through AWS Bedrock.

Prerequisites

Before starting, ensure you have:

• AWS Account: Active AWS account with appropriate permissions
• IAM Permissions: Permissions to invoke Bedrock models (bedrock:InvokeModel or bedrock:Converse)
• AWS Marketplace Permissions: Required for first-time activation of marketplace models

Model Access Overview

Model Access Page Retired

The Model access page in the AWS Bedrock console has been retired. Serverless foundation models are now automatically enabled across all AWS commercial regions when first invoked in your account, so you can start using them instantly without manual activation.

How Model Access Works

Automatic Activation:

• Foundation models are enabled automatically when you first invoke them via API
• No need to manually request access through the console
• Models become available account-wide after first invocation

Special Requirements:

1. Anthropic Models: First-time users must submit use case details before accessing these models (see CLI/SDK method below)
2. AWS Marketplace Models: A user with AWS Marketplace permissions must invoke the model once to enable it account-wide for all users

Access Control:

• Account administrators can control model access through IAM policies
• Use Service Control Policies (SCPs) to restrict access as needed

Getting Started

To start using Bedrock models, you can:

1. Console Method: Select a model from the Model catalog and open it in the playground
2. API Method: Invoke the model using the API operations via CodeMie
3. CLI/SDK Method: Use AWS CLI or SDK to programmatically accept model agreements (required for third-party models like Anthropic)

Recommended Models for AI/Run CodeMie:

Provider	Model	Use Case
Anthropic	Claude 4.5 Sonnet	Primary chat, reasoning, code generation
Anthropic	Claude 4.5 Opus	Complex reasoning, large context tasks
Amazon	Titan Embeddings	Text embeddings for RAG

Regional Considerations

Cross-Region Model Availability

Different models are available in different AWS regions. Not all foundation models are accessible in every region.

Before using models:

1. Verify model availability in your target region
2. Check AWS Bedrock Model Availability
3. Consider data residency and compliance requirements

Quota and Limits

Understanding Quotas

AWS Bedrock has quotas for:

• Requests per minute: Limits on API call frequency
• Tokens per minute: Limits on token processing
• Model-specific limits: Vary by model and region

Requesting Quota Increases

1. Go to AWS Service Quotas console
2. Search for "Bedrock"
3. Select the quota you need to increase
4. Click Request quota increase
5. Provide justification and desired limit
6. Submit request (typically approved within 2-7 days)

Manage Model Access Using SDK and CLI

Programmatic Access

For automation and infrastructure-as-code workflows, you can manage AWS Bedrock model access programmatically using the AWS CLI and SDK. This approach is recommended for production deployments and CI/CD pipelines.

The original AWS documentation can be found here: Access Amazon Bedrock foundation models

Overview

Access to all Amazon Bedrock foundation models is enabled by default with the correct AWS Marketplace permissions in all commercial AWS regions. For programmatic access to third-party models (e.g., Anthropic Claude models), you need to accept the model agreement offers.

Prerequisites for CLI Access

• AWS CLI: Install and configure the latest AWS CLI with Bedrock support
• AWS Credentials: Valid AWS credentials with appropriate IAM permissions
• Bedrock Permissions: IAM permissions for bedrock:* operations

Step 1: List Model IDs for Required Foundation Models

To get the list of available foundation models, run:

aws bedrock list-foundation-models

To filter by a specific provider (e.g., Anthropic):

aws bedrock list-foundation-models --by-provider anthropic | grep modelId

Output example:

"modelId": "anthropic.claude-sonnet-4-20250514-v1:0",
"modelId": "anthropic.claude-haiku-4-5-20251001-v1:0",
"modelId": "anthropic.claude-sonnet-4-5-20250929-v1:0",
"modelId": "anthropic.claude-opus-4-1-20250805-v1:0",
"modelId": "anthropic.claude-opus-4-5-20251101-v1:0",
"modelId": "anthropic.claude-opus-4-20250514-v1:0",

Step 2: List Foundation Model Agreement Offers

Run the following command to list the agreement offers for a specific model (example: Anthropic Claude Haiku 4.5):

aws bedrock list-foundation-model-agreement-offers --model-id anthropic.claude-haiku-4-5-20251001-v1:0

Output example:

{
  "modelId": "anthropic.claude-haiku-4-5-20251001-v1:0",
  "offers": [
    {
      "offerId": "offer-fu6wq1phlosn4",
      "offerToken": "token",
      "termDetails": {
        "usageBasedPricingTerm": {
          "rateCard": [
            {
              "dimension": "USE1_InputTokenCount",
              "price": "1.1",
              "description": "Million Input Tokens Regional",
              "unit": "Units"
            },
            {
              "dimension": "APN3_CacheReadInputTokenCount",
              "price": "0.11",
              "description": "Million Cache Read Input Tokens Regional",
              "unit": "Units"
            }
          ]
        }
      }
    }
  ]
}

Save the Offer Token

Save the offerToken value from the output. You'll need it in Step 4 to create the model agreement.

Step 3: Put Use Case for First-Time Users (Anthropic Models Only)

One-Time Requirement

This step is required only once for Anthropic models and only for first-time users. You can skip this step if you've already completed it for any Anthropic model.

Create a JSON file company.info.json with your company information:

{
  "companyName": "Company XYZ",
  "companyWebsite": "www.xyz.com",
  "intendedUsers": "1",
  "industryOption": "Healthcare",
  "otherIndustryOption": "Life Sciences",
  "useCases": "Agentic AI SDLC tooling"
}

Run the following command:

aws bedrock put-use-case-for-model-access --form-data "$(cat company.info.json | base64)"

Step 4: Create Foundation Model Agreement

Create the agreement (access) for the foundation model using the offer token and model ID from the previous steps:

aws bedrock create-foundation-model-agreement --model-id anthropic.claude-haiku-4-5-20251001-v1:0 --offer-token "token"

Output example:

{
  "modelId": "anthropic.claude-haiku-4-5-20251001-v1:0"
}

Step 5: Get Foundation Model Availability

Check if the model is available for use:

aws bedrock get-foundation-model-availability --model-id anthropic.claude-haiku-4-5-20251001-v1:0

Output example:

{
  "modelId": "anthropic.claude-haiku-4-5-20251001-v1",
  "agreementAvailability": {
    "status": "PENDING"
  },
  "authorizationStatus": "AUTHORIZED",
  "entitlementAvailability": "AVAILABLE",
  "regionAvailability": "AVAILABLE"
}

Wait for Activation

The agreementAvailability.status will initially show PENDING. Wait patiently until it changes to AVAILABLE. This may take several minutes for first-time users.

You can poll the status by running the same command periodically:

# Check status every 30 seconds
watch -n 30 'aws bedrock get-foundation-model-availability --model-id anthropic.claude-haiku-4-5-20251001-v1:0'

Step 6: Activate All Other Required Foundation Models

Repeat Steps 2, 4, and 5 for all other foundation models you need to activate:

1. List agreement offers (Step 2)
2. Create model agreement (Step 4)
3. Verify availability (Step 5)

Batch Processing

For multiple models, consider creating a shell script that loops through your required model IDs to automate the activation process.

Step 7 (Optional): Delete Foundation Model Agreement

If you need to revoke access to a model, run:

aws bedrock delete-foundation-model-agreement --model-id anthropic.claude-haiku-4-5-20251001-v1:0

Access Revocation

This command will revoke your access to the model. The agreementAvailability.status will change to NOT_AVAILABLE when you check with get-foundation-model-availability.

Next Steps

After enabling model access, configure AI/Run CodeMie to use these models by updating the AI models integration configuration. Refer to the AI Models Integration Overview for detailed configuration instructions.