Accelerate productive AI production in Canada with Amazon Bedrock Cross-Region Cross-Region

Courtnative AI has created unprecedented opportunities for Canadian organizations to transform their operations and customer experience. We are pleased to announce that customers in Canada can now access advanced base models including Anthropic's Sonnet Claude 4.5 and Claude Haiku 4.5 at Amazon Bedrock through Cross-Region Reference (CRIS).

This post identifies how Canadian organizations can use regional attack profiles in the Canadian region (Central) to access the latest baseline models to accelerate AI efforts. We will show how to get started with these new capabilities, provide a migration guide from old models, and share recommended practices for quota management.

Canadian Cross-Reciount -UCK WECONEL: Your Gateway to Global Ai Innovation

To help customers scale their AI production systems, Amazon Bedrock offers Cross-Region Supcer profiles (CRIS).

Amazon Bedrock offers two types of regional elevation profiles

1. Geographic Cris: Amazon Bedrock automatically selects the appropriate trading region within that geography to process your promotion request.
2. Global Cris: Global Cris also improves region-to-region adoption by enabling the routing of lubricant applications to commercial platforms designed for global sales, optimizing available resources and enabling a superior model.

Regional authorization works by using a secure AWS network with End-to-End Becctionion of both data in transit and at rest. When a customer applies for a regional upgrade from the Canadian (Central) region, CRIS directs the request to one of the destinations configured for the renewal profile (US or international profiles).

The key distinction is that while inference processing (the transient computation) may occur in another Region, all data at rest—including logs, knowledge bases, and any stored configurations—remains exclusively within the Canada (Central) Region. The upload request travels over the AWS Global network, which it broadcasts to create the Public Internet, and the responses are returned encrypted to your application in Canada.

Regional configuration in the chest of Canada

Through CRIS, Canadian organizations get early access to Foundation models, including cutting-edge models like Claude Sonnet 4.5 with advanced development, providing a faster way to innovate. CRIS also delivers improved power and efficiency by providing power access to multiple regions. This enables higher throughput during peak times such as tax time, Black Friday, and holiday shopping, automated management without manual intervention, and increased scalability by serving requests from a larger pool of resources.

Canadian customers can choose between two profile types based on their needs:

Starting with Cris from Canada

To start using regional adoption from Canada, follow these steps:

Configure AWS identity and access management (IAM)

First, verify your IAM role or user has the necessary permissions to request Amazon Bedrock modules that use regional elevation profiles.

Here is an example of a US state naming policy:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel*"
            ],
            "Resource": [
                "arn:aws:bedrock:ca-central-1::inference-profile/us.anthropic.claude-sonnet-4-5-20250929-v1:0"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "bedrock:InvokeModel*"
            ],
            "Resource": [
                "arn:aws:bedrock:*::foundation-model/anthropic.claude-sonnet-4-5-20250929-v1:0"
            ],
            "Condition": {
                "StringLike": {
                    "bedrock:InferenceProfileArn": "arn:aws:bedrock:ca-central-1::inference-profile/us.anthropic.claude-sonnet-4-5-20250929-v1:0"
                }
            }
        }
    ]
}

For Global Cris refer to the Blog post, Open Global Ai Infence Scalability using new Global Cross-Region cross detection in the Amazon region with Anthropic's Claude Sonnet 4.5.

Use regional lift profiles

Configure your application to use the appropriate profile ID. Profiles use prefixes to indicate their route width:

Example code

Here's how to use Amazon Bedrock API US Cris Tris Onence from Canada:

import boto3

# Initialize Bedrock Runtime client
bedrock_runtime = boto3.client(
    service_name="bedrock-runtime",
    region_name="ca-central-1"  # Canada (Central) Region
)

# Define the inference profile ID
inference_profile_id = "us.anthropic.claude-sonnet-4-5-20250929-v1:0"

# Prepare the conversation
response = bedrock_runtime.converse(
    modelId=inference_profile_id,
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "text": "What are the benefits of using Amazon Bedrock for Canadian organizations?"
                }
            ]
        }
    ],
    inferenceConfig={
        "maxTokens": 512,
        "temperature": 0.7
    }
)

# Print the response
print(f"Response: {response['output']['message']['content'][0]['text']}")

Quota management of Canadian workloads

When using CRIS from Canada, quota management is done at the regional resource level (ca-central-1). This means the quota increase requested by the Canadian region (Central)) applies to all quota requests from Canada, regardless of where they are processed.

Calculating Quota Statistics

IMPORTANT: When calculating your required rate increase, you need to look at the burn rate, defined as the rate at which input and output tokens are converted into Token Quota Use of the throwing system. The following models have a 5x hot down rate for exit tokens (1 Exit token consumes 5 tokens from your Quotas):

• Anthropic Claude Opus 4
• Anthropic Claude Sonnet 4.5
• Anthropic Claude Sonnet 4
• Anthropic Claude 3.7 Sonnet

In other models, the burn rate is 1:1 (1 token withdrawal uses 1 token from your Quota). To enter tokens, the token to quota ratio is 1: 1. The calculation of the total number of tokens per request is as follows:

Input token count + Cache write input tokens + (Output token count x Burndown rate)

Asking for a rate increase

To request a quota increase through CRIS Canada:

1. Navigate to the AWS Service Quote Console for the Canada (Central) region
2. Search for a specific model rate (for example, “Claude Sonnet tokens 4.5 per minute”)
3. Apply for an increase based on your verified usage

Migration from old Claude models to Claude 4.5

Organizations currently using older Claude models should plan their migration to Claude 4.5 to ensure the latest capabilities of the model.

To plan your migration strategy, include the following items:

1. Benchmark performance current performance: Establish basic metrics with your existing models.
2. Check with the workloads you represent and upgrade: Verify the performance of Claude 4.5 for your specific use cases, then quickly adjust the optimization capabilities of Claude 4.5 and use the Bedrock Prompt Optimizer tool.
3. Start working slowly: Slowly converting traffic.
4. Monitor and change: Track performance metrics and adjust measurements as needed.

Choose between US and Global profiles

When using CRIS from Canada, organizations can choose between US and Globalfence profiles based on their specific needs.

US regional importation is recommended for organizations with existing US data agreements, high throughput and robustness requirements and development and testing facilities.

Lasting

The region's discovery of the Amazon Bedrock region represents an opportunity for Canadian organizations looking to use AI while maintaining data governance. By distinguishing between continuous neutrality operations and continuous data storage, CRISs provide fast access to the latest base models without compromising compliance requirements.

With CRIS, Canadian organizations gain access to new models within days instead of months. The system scales automatically during peak business hours while maintaining full audit trails within Canada. This helps you meet compliance requirements and use the same advanced AI capabilities as organizations around the world. To get started, review your data management requirements and configure IAM permissions. Then experiment with the compatibility profile that suits your needs – with us for low latency in the US states, or around the world for maximum power.

About the writers

Daniel Duplessis He is the lead AI asset architect at Amazon Web Services (AWS), where he guides businesses in understanding AI implementation strategies and develops core AI support capabilities.

Dan mackay He is a financial services compliance specialist at AWS Canada. He advises clients on best practices and effective solutions for cloud-related governance, risk and compliance. Dan looks to help AWS customers navigate the financial services and privacy laws that apply to the use of cloud technology in Canada with a focus on external risk and operational resilience.

Melanie isPhD, is a specialized AI solutions specialist based in Sydney, Australia, where his focus is working with clients to build solutions using AI / ML tools. He has been actively involved in many AI production projects at APJ, combining the strengths of LLMS. Prior to joining AWS, Dr. Li held data science roles in the financial and retail industries.

Serge Malikov is a senior solution manager based in Canada. His focus is on the financial services industry.

Saurabh trikande Senior product manager for Amazon Bedrock and Amazon SageMaker Softening. He is passionate about working with customers and partners, motivated by the goal of Demomentatizeng AI. He focuses on the key challenges related to implementing complex AI applications, discovery through multivariate models, costing, and making the deployment of productive AI models accessible. In his spare time, Saurabh enjoys hiking, learning about new technologies, following techcrunks, and spending time with his family.

Sharadha Kandasubramanian Senior technical program manager for Amazon Bedrock. He drives Genai's active programs for Amazon Bedrock, allowing customers to grow and scale their genai operation. Outside of work, he is an avid runner and biker who enjoys spending time outside in the sun.