Introducing cross-Region inference for OpenAI GPT-5.6 models on Amazon Bedrock
Amazon Bedrock now offers OpenAI GPT-5.6 models (Sol, Terra, and Luna) in more than 25 AWS Regions with cross-Region inference. Learn how US geographic and global inference profiles route requests for higher throughput, how to call the models with the OpenAI and Converse APIs, and how to configure IAM, quotas, and monitoring.
OpenAI and Amazon collaborate to provide GPT-5.6 models on Amazon Bedrock across over 25 AWS Regions, introducing cross-Region inference. Three GPT-5.6 variants, Sol, Terra, and Luna, each offer a unique balance of capability and cost for cross-Region inference. Cross-Region inference (CRIS) is achieved through inference profiles, which define a model and the AWS Regions to which requests are routed.
CRIS improves throughput and maintains consistent performance under load by enabling requests to draw on a broader pool of compute. Geographic CRIS routes requests within a single geography, ensuring data stays within that geography, while global CRIS routes across all supported AWS commercial Regions based on real-time capacity.
The launch introduces US and global geographic CRIS for GPT-5.6, with US geographic CRIS covering US Regions and global CRIS covering all supported commercial Regions. The three GPT-5.6 variants - Sol, Terra, and Luna - accept text and image inputs, return text, and support reasoning mode, server-side tool calling, and prompt caching.
They can be called using the OpenAI Responses API, OpenAI Chat Completions API, and Amazon Bedrock Converse API, with streaming support through the Responses and Chat Completions API (stream=True) and ConverseStream. Geographic and global inference profiles are logical identifiers passed instead of a raw model ID, with geographic profiles keeping inference processing within predefined geographies and global profiles routing requests based on real-time capacity.
Billing and quota consumption are tracked regardless of which backend Region handled the request. Data processed through global CRIS may cross Regions in the model's eligible set, so geographic profiles or direct calls to specific Regions are recommended for workloads with data residency requirements.
Written by urgent.news from AWS Machine Learning's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
