Access OpenAI GPT models on Amazon Bedrock from Canada with cross-Region inference
aws ·
AWS has enabled Canadian organizations to access OpenAI GPT-5.6, GPT-6, and GPT-6.1 models via Amazon Bedrock using cross-Region inference (CRIS). This feature allows requests originating from the Canada (Central) and Canada West (Calgary) Regions to be routed to US or other global regions for processing, thereby expanding capacity during demand spikes. Users can choose between 'US geographic' profiles, which keep inference within North America, and 'Global' profiles that draw from a broader capacity pool. The article provides detailed guidance on setting up IAM permissions, configuring Service Control Policies for data residency, and invoking models through the OpenAI Responses API, Chat Completions API, and Amazon Bedrock Converse API. It also covers practical aspects such as prompt caching for cost reduction, quota management for tokens per minute, and observability tools like CloudTrail and CloudWatch. Security measures ensure data remains encrypted in transit and at rest in the source Region, with specific notes on content moderation retention in destination regions. Finally, the post outlines steps for monitoring usage, requesting quota increases, and cleaning up test resources to avoid unnecessary charges.
aws