Artificial Intelligence Australian teams working with OpenAI models can now access the latest OpenAI models through Amazon Bedrock. Amazon Bedrock offers OpenAI GPT-5.6 Sol, Terra, and Luna with global cross-Region inference from both Asia Pacific (Sydney) and Asia Pacific (Melbourne) AWS Regions in Australia. Your application calls the Amazon Bedrock Runtime endpoint in Asia Pacific (Sydney) or Asia Pacific (Melbourne), and Amazon Bedrock routes the request to a supported commercial AWS Region for processing. This provides access to a broader capacity pool without requiring applications to manage destination Region routing. GPT-5.6 Sol is suited to demanding reasoning, coding, and agentic workloads. Terra balances performance and cost for everyday production use. Luna provides fast, affordable inference for high-volume and latency-sensitive applications. All three models accept text and image inputs, generate text, and support context windows of up to 1 million tokens. With the Amazon Bedrock Runtime endpoint, you can invoke these models with Responses API, Chat Completions API, and the Converse API from Asia Pacific (Sydney) and Asia Pacific (Melbourne).
In this post, we show you how to use prompt caching to optimize your inference cost, set up Codex with OpenID Connect (OIDC)-based authentication, and monitor usage with Amazon CloudWatch and Coding Agent Insights.
GPT-5.6 global inference profiles #
The following table lists the three global profile IDs and the Australian source Regions covered in this post.
Global inference profiles | Source Regions | Destination Regions | | global.openai.gpt-5.6-sol global.openai.gpt-5.6-terra global.openai.gpt-5.6-luna | Asia Pacific (Sydney) ap-southeast-2 Asia Pacific (Melbourne) ap-southeast-4 | Routes to supported commercial AWS Regions |
Profile membership and model availability can change. Check Cross-Region inference support before deployment.
Getting started #
Before proceeding with this section, you can follow the GPT-5.6 blog post to set up the following prerequisites:
- An AWS account with Asia Pacific (Sydney) or Asia Pacific (Melbourne) enabled as your source Region.
- If your organization uses service control policy (SCP), verify that the SCP allows the GPT-5.6 global inference profiles in the selected source Region.
- An AWS Identity and Access Management (IAM) role or user with proper permissions to invoke the GPT-5.6 inference profiles.
- Python 3.9 or later with the
openai
,boto3
, andaws-bedrock-token-generator
packages installed.
In the following section, we show you how to verify the active global inference profiles with both the AWS Command Line Interface (AWS CLI) and the Amazon Bedrock console.
Using the AWS CLI. The following commands list the active GPT-5.6 profiles and inspect the Terra profile from the Sydney Region. You can also use the commands with Sol or Luna by changing the inference profile ID. To run the same check from Melbourne Region, replace ap-southeast-2
with ap-southeast-4
.
Using the Amazon Bedrock console. Open the Amazon Bedrock console, select Sydney or Melbourne as your Region and choose Inference profiles under Infer, and filter for Global OpenAI GPT-5.6 Terra. The following screenshot shows the active profile from the Sydney source Region.
Invoke GPT-5.6 through Amazon Bedrock Runtime
GPT-5.6 supports three access paths on the Amazon Bedrock Runtime endpoint: the OpenAI Responses API, OpenAI Chat Completions API, and Amazon Bedrock Converse API. The OpenAI-compatible APIs are called on the /openai/v1
paths of this endpoint rather than through the AWS SDKs. The endpoint accepts either AWS Signature Version 4 (SigV4) or an Amazon Bedrock model inference API key. The following example uses the AWS Bedrock Token Generator for Python to create a short-term Amazon Bedrock model inference API key from the current AWS credentials, so the application does not need to store a static key. You can then create an OpenAI client using this short-term API key to work with the supported APIs.
OpenAI Responses API. For applications that already use the OpenAI SDK with the Responses API, you can point the client at the Regional Amazon Bedrock Runtime endpoint.
For streaming output, set stream=True
and iterate over the response events. The following example prints text as it arrives:
OpenAI Chat Completions API. You can also work with the Chat Completions API if your application already uses that.
Amazon Bedrock Converse API. Use Converse API when your application calls Amazon Bedrock through an AWS SDK. Boto3 resolves credentials through the standard AWS credential chain.
For streaming output, use converse_stream
with the same Region and profile ID, then iterate over the returned event stream.
The preceding examples call Amazon Bedrock in Asia Pacific (Sydney). To run the same example for Asia Pacific (Melbourne), set the Region to ap-southeast-4
and run the code.
Working with prompt caching
GPT-5.6 prompt caching is available through the supported APIs. GPT-5.6 supports two caching modes on Amazon Bedrock. Implicit caching is enabled by default, and no code changes are required, while with explicit caching, you can define the reusable prefix, cache boundary, and cache key. The GPT-5.6 blog post provides examples that illustrate the prompt caching capability.
Setting up Codex with GPT-5.6 on Amazon Bedrock #
Codex can use the same global inference profiles through Amazon Bedrock Runtime. Install the latest Codex CLI to use the native Amazon Bedrock Runtime model provider. Here, we validated the following configuration with codex-cli 0.149.1 using GPT-5.6 Sol from Asia Pacific (Sydney).
For organizations whose identity provider is Okta, Auth0, Microsoft Entra ID, Amazon Cognito, or AWS IAM Identity Center, the AWS OIDC Auth Helper repository provides a sample credential helper. First, follow the guide to configure your identity provider, the corresponding AWS federation resource, and an IAM role carrying the Amazon Bedrock permissions shown earlier. Then add a named profile to ~/.aws/config
, so the helper does not replace credentials resolved by the default profile. Replace the placeholders with the path to the installed helper and the profile name defined in the ~/.aws/config
file.
This federation helper exchanges an OIDC token for temporary AWS credentials, and Codex reads through the standard AWS credential chain with no further configuration. Next, create or update ~/.codex/config.toml
and reference the AWS profile, see the Codex configuration reference for other supported settings:
If the helper doesn’t have a valid cached session, it opens the configured sign-in page in your browser. After you authenticate, the helper returns temporary AWS credentials through credential_process
. Requests are signed with AWS SigV4, so no API key is involved in the inference path. When the profile is backed by AWS IAM Identity Center, the credentials are already short-term and rotate with the single sign-on session. To use this with Asia Pacific (Melbourne) Region, set the Region to ap-southeast-4
in the AWS profile and Codex configuration.
Quota management #
GPT-5.6 on-demand quotas are measured in requests per minute (RPM) and tokens per minute (TPM). Token burndown determines how each request consumes TPM. Token consumption is calculated from input tokens, cache-write input tokens, and output tokens multiplied by the model’s burndown rate. For instance, for GPT-5.6, input tokens and cache-write input tokens count at 1:1, while each output token consumes 10 tokens from the quota. Review the GPT-5.6 quotas in the Service Quotas console from the source Region your application uses: Asia Pacific (Sydney) Region (ap-southeast-2
) or Asia Pacific (Melbourne) Region (ap-southeast-4
). Before production rollout, request increases early, monitor quota utilization, and test representative prompts, output lengths, streaming behavior, concurrency, and peak traffic. See Amazon Bedrock quotas for the current values and token burndown rates.
Monitoring and logging #
Because GPT-5.6 requests use the Amazon Bedrock Runtime API, requests made through the global inference profiles appear in model invocation logging like other on-demand requests. When logging is enabled, records include the model or inference profile ID used for the call and invocation metadata. Codex uses OpenTelemetry (OTel) and exports metrics over OTLP/HTTP, see set up OpenTelemetry for OpenAI Codex for more details.
CloudWatch Coding Agent Insights provides a dashboard for Codex telemetry, including token usage, API requests, active users, conversation activity, and optional organizational dimensions. There are two paths to configure Coding Agent Insights in CloudWatch for Codex, that is, using Bearer token or Enterprise rollout.
The following example shows how to configure the Coding Agent Insights for Codex using the Bearer approach. First, get a CloudWatch metrics API key, then add the following sections to ~/.codex/config.toml
.
Replace YOUR_CLOUDWATCH_METRICS_API_KEY
with the key created in CloudWatch, then start Codex. This CloudWatch metrics API key can then authorize the export to the Regional CloudWatch endpoint. After telemetry arrives, open the CloudWatch console in Asia Pacific (Sydney) Region, choose GenAI Observability, Coding Agent Insights, and the Codex tab. You will see the Coding Agent Insights dashboard displays Codex token usage, request activity, cache hit rate, and more.
To populate the Organization, Environment, Department, Cost Center, Location, Team, and User filters, provide the corresponding values through OTEL_RESOURCE_ATTRIBUTES
. AWS classifies that CloudWatch metric API key as a long-term credential and recommends it only where short-term AWS credentials are not feasible. Treat config.toml
as a secret and restrict its file permissions.
For organizations that federate developer identity through corporate single sign-on, we recommend using the enterprise rollout, where a local collector signs the export with SigV4 using the developer’s federated credentials and no token is distributed.
Conclusion #
In this post, we showed how to discover and invoke the GPT-5.6 Sol, Terra, and Luna global inference profiles from Asia Pacific (Sydney) and Asia Pacific (Melbourne) Regions. We also introduced how to configure Codex to use Amazon Bedrock Runtime and export Codex telemetry to CloudWatch Coding Agent Insights.
To get started, follow the examples in this post and test GPT-5.6 models in your account. If you are using Codex, you can configure the Amazon Bedrock Runtime provider and enable CloudWatch Coding Agent Insights in the AWS Console to monitor your Codex consumption. For pricing details, see Amazon Bedrock pricing.