OutYet reporting
AWS documents a new Bedrock route for GPT-5.6 Sol in Australia
AWS has documented a cross-Region Bedrock path for GPT-5.6 Sol from Sydney and Melbourne, changing routing and deployment constraints rather than the model itself.
Amazon Web Services says developers can invoke GPT-5.6 Sol from its Sydney and Melbourne Bedrock regions through `global.openai.gpt-5.6-sol`, a global cross-Region inference profile. The application calls the Bedrock Runtime endpoint in Australia while Bedrock selects a supported commercial destination region for processing. This is an access and routing change for Australian AWS users, not evidence of a different Sol capability tier or a new model identifier.
The Australian guidance is the regional follow-through to AWS's August expansion of GPT-5.6 support on the Bedrock Runtime endpoint, which added Responses, Converse, and Chat Completions interfaces alongside geographic and global routing. AWS's broader technical guide describes the global profile as a capacity mechanism: requests draw on a wider pool than a single region, while the caller uses one endpoint and one profile ID. In that context, the September post makes Sydney and Melbourne explicit source regions rather than changing the underlying GPT-5.6 family.
The practical distinction is between Bedrock's two access paths. The Sol model card lists the raw `openai.gpt-5.6-sol` identifier for the `bedrock-mantle` endpoint, but the newer `bedrock-runtime` path requires an inference-profile identifier such as `global.openai.gpt-5.6-sol` and does not present Sol as an in-Region runtime choice. A migration is therefore more than a base-URL swap: the model parameter and permissions must match the selected endpoint and profile.
The global option also has a boundary. AWS says it may route inference to any supported commercial region, so it is a poor fit where a workload's policy requires Australian-only processing or a fixed data-processing geography. AWS also says profile membership and availability can change, and advises checking Cross-Region Inference support before deployment. Teams should validate residency rules, IAM policy, capacity behavior, and the current pricing table in their own account before standardizing on the profile.
Related models
Sources
- Accessing OpenAI models on Amazon Bedrock from Australia with global cross-Region inference · Amazon Web Services
- GPT-5.6 Sol - Amazon Bedrock · Amazon Web Services
- Introducing cross-Region inference for OpenAI GPT-5.6 models on Amazon Bedrock · Amazon Web Services