Accessing OpenAI GPT models on Amazon Bedrock with global cross-Region inference in Thailand, Malaysia, Singapore, Indonesia, and Taiwan
aws ·
AWS has announced global cross-Region inference capabilities for OpenAI GPT-6 models on Amazon Bedrock, specifically targeting users in Thailand, Malaysia, Singapore, Indonesia, and Taiwan. This feature allows teams to access GPT-6 Sol, GPT-6 Luna, and GPT-6 Astra from five Asia Pacific source regions by routing requests to broader commercial AWS Regions when local capacity is insufficient. The article provides technical guidance on configuring these profiles via the Converse API, OpenAI-compatible APIs, and coding clients like Codex. It details necessary IAM policies and Service Control Policy adjustments to handle the 'unspecified' region attribute inherent in cross-Region calls. Quota management is tracked in the source region using a token-per-minute model that accounts for input, cache, and output tokens. Monitoring and billing remain centralized in the source region, with CloudTrail logs indicating the specific destination region for each inference request. The post concludes with security considerations regarding data retention and steps for cleaning up temporary configurations.
aws