AWS Bedrock Adds Cross-Region Inference for OpenAI's GPT-5.6 Models
Amazon Bedrock now supports cross-Region inference for three GPT-5.6 variants — Sol, Terra, and Luna — across more than 25 AWS Regions. The feature routes requests to available compute capacity via geographic or global inference profiles, without requiring code changes beyond swapping a model ID.
Amazon Bedrock now supports cross-Region inference (CRIS) for OpenAI's GPT-5.6 models, according to a joint post from AWS and OpenAI's Chris Dickens. The feature is live across more than 25 AWS Regions and applies to three GPT-5.6 variants — Sol, Terra, and Luna — each described as offering a different balance of capability and cost.
What cross-Region inference does
CRIS is a capacity and routing mechanism, not a new model. It works through inference profiles — logical identifiers passed in place of a raw model ID — that tell Bedrock which Regions can handle a given request. Two profile types are now available for GPT-5.6:
- Geographic inference profile (e.g.,
us.openai.gpt-5.6-terra): routes requests only within a defined geography, such as the US and Canada, keeping data processing inside that boundary for residency-sensitive workloads. - Global inference profile (e.g.,
global.openai.gpt-5.6-terra): routes requests to any supported commercial AWS Region where the model is deployed, based on real-time capacity, offering the largest available compute pool.
AWS states that billing and quota consumption are tracked against the calling account regardless of which backend Region actually processes the request, so customers see a single spending and throughput view. The company notes that global CRIS may route data across Regions in the model's eligible set, and recommends geographic profiles or direct single-Region calls for workloads with strict data residency requirements.
Model specs and access
All three GPT-5.6 variants covered by this launch — Sol, Terra, and Luna — accept text and image input, return text, and support a 1 million token context window, according to AWS. They also support reasoning mode, server-side tool calling, and prompt caching. A separate cybersecurity-specialized variant of GPT-5.6 exists but is not covered by this cross-Region inference update.
The models can be called through the OpenAI Responses API, the OpenAI Chat Completions API, or the Amazon Bedrock Converse API, with streaming supported via stream=True on the OpenAI-compatible endpoints and via ConverseStream on Bedrock's native API. AWS says existing OpenAI SDK clients can point at Bedrock's OpenAI-compatible endpoint and swap in the geographic or global inference profile ID as the model parameter, using either standard AWS credentials or a short-term Bedrock API key generated via the aws-bedrock-token-generator package.
US source Regions (N. Virginia, Ohio, Oregon, N. California, Canada Central, Canada West) route within a US/Canada geographic pool. The global profile spans source Regions across the US, Canada, Europe, Asia Pacific, the Middle East, and South America, routing to any supported commercial Region worldwide.
Pricing for GPT-5.6 on Bedrock was not disclosed in this announcement.
What this means
This is an infrastructure and availability update, not a new model launch — GPT-5.6 itself is not introduced here, only its routing options on Bedrock. The practical value is capacity resilience: enterprises running high-volume inference no longer need to provision against a single Region's throughput ceiling, and can choose between strict data-residency compliance (geographic profiles) and maximum capacity (global profiles). This mirrors AWS's existing cross-Region inference pattern already used for other foundation models on Bedrock, extending it to OpenAI's GPT-5.6 family as part of the broader OpenAI-on-Bedrock partnership. For teams already running OpenAI SDK-based applications, the migration path — swapping a model ID for an inference profile ID — is deliberately low-friction.
Related Articles
Kimi Code CLI v2.7.0 adds native Amazon Bedrock provider with standard AWS credentials
Kimi Code CLI v2.7.0, Moonshot AI's open source coding agent, adds a native bedrock provider. It resolves the standard AWS credential chain, including named profiles and SSO, so users no longer need a separate Amazon Bedrock API key. AWS demonstrates the setup with Kimi K3.1 Pro Highspeed.
OpenAI tells developers to rewrite system prompts for GPT-6.2 Astra and Sol as defaults shift
OpenAI says developers moving to GPT-6.2 Astra or Sol should review their system prompts. The company published copy-ready templates in the models' documentation, and the Responses API now discards earlier reasoning content if conversation history is edited.
OpenAI releases GPT-Live-1.1 in the API, with standby mode billing idle WebRTC time at $0.02 per minute
OpenAI has released GPT-Live-1.1 in the API. The full-duplex voice model improves multilingual accents and transcription, and a new standby mode bills idle WebRTC time at $0.02 per minute instead of $0.05.
GPT-6.2 Astra scores 41.2% on XDOF's Ember-0 robot benchmark, up from 19.6% for GPT-6 Astra
Robotics infrastructure company XDOF reports that GPT-6.2 Astra reached a 41.2% average success rate across the 16 tasks of its Ember-0 benchmark, versus 19.6% for GPT-6 Astra. It is the first model to complete rope knot-tying and T-shirt unfolding on the benchmark. Average prediction latency fell from 15 seconds to 10.1 seconds.
Comments
Loading...