OpenAI GPT-6 models now reachable from five Asian AWS Regions via Bedrock

How global cross-Region inference works

AWS says single-Region inference can throttle when demand outstrips local capacity. Global cross-Region inference is its answer: requests are routed across supported commercial AWS Regions where the model is deployed, which gives callers broad access to Amazon Bedrock capacity.

Developers use it by passing a global cross-Region inference profile ID in place of a model ID. The source Region is the one where the API call is made, and Amazon Bedrock picks the destination Region for processing. According to AWS, billing, quota tracking, AWS CloudTrail and Amazon CloudWatch all stay in the source Region.

The profile IDs are global.openai.gpt-6-sol, global.openai.gpt-6-luna and global.openai.gpt-6-astra. Sol and Luna can be called from Singapore (ap-southeast-1), Jakarta (ap-southeast-3), Malaysia (ap-southeast-5), Thailand (ap-southeast-7) and Taipei (ap-east-2). Astra is listed only for Singapore. AWS advises checking the model cards before building, because profile availability and Region support change as AWS adds Regions and profiles.

The three GPT-6 models

AWS describes GPT-6 Sol as bringing strong reasoning and coding capabilities to complex tasks performed throughout the week. GPT-6 Luna, it says, makes focused, repeatable tasks practical at high volume. GPT-6 Astra handles complex reasoning, coding and long-document work.

All three accept text and image input, client-side tool use and prompt caching. They differ in reasoning effort settings:

• GPT-6 Sol and GPT-6 Luna accept none, low, medium, high, xhigh and max.
• GPT-6 Astra accepts the same levels except none.

For per-token rates, AWS points readers to the Amazon Bedrock pricing page. The post gives no price figures.

Getting started

The prerequisites are an IAM identity with permission to invoke the profile and an activated source Region. Destination Regions need no activation, since they run on Amazon Bedrock managed capacity. Singapore is enabled by default, while Jakarta, Malaysia, Thailand and Taipei are opt-in Regions that must be activated first.

Each model supports three access paths on the bedrock-runtime endpoint: the Amazon Bedrock Converse API, the OpenAI Responses API and the OpenAI Chat Completions API. The OpenAI-compatible APIs are called on the /openai/v1 paths rather than through the AWS SDKs. The endpoint accepts either Signature Version 4 (SigV4) or an Amazon Bedrock API key. The post includes Python examples for the Converse API and the Chat Completions API, the latter using the OpenAI SDK and the Amazon Bedrock token generator.

For teams migrating an existing OpenAI client, AWS lists several requirements: set the model parameter on every request, use the supported tool types (function, mcp, custom, namespace and tool_search), and configure the inference profile ID directly in the client rather than querying the models endpoint.

ChatGPT desktop app and Codex CLI

The ChatGPT desktop app and the Codex CLI share the ~/.codex/config.toml file, so a single change covers both. Of the two built-in Amazon Bedrock providers, only amazon-bedrock-runtime accepts an inference profile. AWS shows a configuration that sets the model to global.openai.gpt-6-astra and the region to ap-southeast-1.

Because requests are signed with SigV4, AWS notes there is no key to rotate. Codex reports a 272,000-token window for these profiles, which AWS describes as Codex's own bundled figure rather than a service limit, and it can be raised with model_context_window. The configuration needs Codex CLI 0.148 or later.

What it means for developers in the region

For teams in Thailand, Malaysia, Singapore, Indonesia and Taiwan, the change offers a way to reach the newest OpenAI models without depending on local capacity alone, while keeping billing, quotas and logging in their own Region. The post also covers IAM and service control policy configuration, quota management and monitoring, though the source text available here is cut off before those details.