CONFIRMED: AWS set out the access in a Machine Learning Blog post on 12 October 2026, co-written with OpenAI.
Developers in Thailand, Malaysia, Singapore, Indonesia and Taiwan can now call OpenAI’s GPT-6 models on Amazon Bedrock without being held back by a single Region’s capacity. AWS says “Amazon Bedrock offers OpenAI GPT-6 Sol and GPT-6 Luna through global cross-Region inference from all five AWS Regions, and GPT-6 Astra from Asia Pacific (Singapore)” [1].
This is not GPT-6 arriving on Bedrock for the first time. GPT-6 Astra was already available there. What is new is the global routing from these source Regions.
Cross-Region inference means your request is sent from the Region you call (the “source Region”) and Amazon picks another Region with spare capacity to process it. AWS explains why: “Single-Region inference can throttle when demand exceeds local capacity” [1].
Which models, from where
| Profile ID | Source Regions |
|---|---|
global.openai.gpt-6-sol and global.openai.gpt-6-luna | Singapore (ap-southeast-1), Jakarta (ap-southeast-3), Malaysia (ap-southeast-5), Thailand (ap-southeast-7), Taipei (ap-east-2) |
global.openai.gpt-6-astra | Singapore (ap-southeast-1) |
For destinations, AWS lists only “Supported commercial AWS Regions where the model is deployed” [1]. You use a profile by passing its ID “in place of a model ID” [1].
Singapore is on by default. AWS says Jakarta, Malaysia, Thailand and Taipei “are opt-in Regions you must activate first” [1].
What the models do
AWS describes the three models this way [1]:
- GPT-6 Sol “brings strong reasoning and coding capabilities to complex tasks performed throughout the week.”
- GPT-6 Luna “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” [1].
Reasoning effort. This setting controls how hard the model “thinks” before it answers. “GPT-6 Sol and GPT-6 Luna accept none, low, medium, high, xhigh, and max. GPT-6 Astra accepts the same levels except none” [1].
Two ways to call them
The bedrock-runtime endpoint offers three routes for each model: “the Amazon Bedrock Converse API, the OpenAI Responses API, and the OpenAI Chat Completions API” [1].
- Converse API: Amazon’s own interface, called through the AWS SDKs.
- OpenAI-compatible APIs: these “are called on the
/openai/v1paths rather than through the AWS SDKs, and the endpoint accepts either Signature Version 4 (SigV4) or an Amazon Bedrock API key” [1].
So, AWS says, “an existing OpenAI integration moves across by changing a base URL and a key” [1].
The post also covers the ChatGPT desktop app and Codex CLI. AWS notes that “Codex reports a 272,000-token window for these profiles, which is its own bundled figure rather than a service limit” [1]. That setup needs Codex CLI 0.148 or later.
Billing and monitoring stay at home
AWS says: “Billing, quota tracking, AWS CloudTrail, and Amazon CloudWatch all stay in your source Region” [1]. It adds that customers pay “that Region’s rates, not the processing Region’s” [1]. CloudTrail records where each request was actually processed, in a field called additionalEventData.inferenceRegion [1].
Quotas. Quotas are counted in tokens per minute. “Input and cache-write tokens count 1:1, and each output token consumes 10 tokens of quota” [1].
Data and compliance
AWS says traffic flagged by its abuse classifiers “is retained by AWS for up to 30 days”. Under cross-Region inference, that retention “happens in the destination Region” [1]. Its advice to organisations with data-residency rules: “take this to your risk and compliance team early” [1].
AWS did not give prices in the post. It points readers to the Amazon Bedrock pricing page.
Sources
- AWS Machine Learning Blog (Traci Lim, Chris Dickens, Karan Jhaveri, Melanie Li, Saurabh Trikande, Zohreh Norouzi), “Accessing OpenAI GPT models on Amazon Bedrock with global cross-Region inference in Thailand, Malaysia, Singapore, Indonesia, and Taiwan”, 12 October 2026. https://aws.amazon.com/blogs/machine-learning/accessing-openai-gpt-models-on-amazon-bedrock-with-global-cross-region-inference-in-thailand-malaysia-singapore-indonesia-and-taiwan/
