OpenAI’s GPT-5.6 Sol, Terra, and Luna models are now generally available on Amazon Bedrock. The post explains how customers can choose a model, run inference through the Responses API on the bedrock-mantle endpoint, and use prompt caching to reduce cost.
The availability matters for companies that already use Bedrock to manage model access, permissions, billing, and deployment patterns. Adding OpenAI models gives those teams another option without moving every workflow to a separate provider interface.
AWS also describes connecting the OpenAI Codex coding agent and planning for quotas and scaling. Those details are important because model access alone is not enough for production use; teams need predictable throughput, cost controls, and integration points for existing developer tools.
The post is a getting-started guide, not an independent performance review. Customers still need to test the models against their own accuracy, latency, safety, and price requirements before replacing an existing Bedrock model in a live application.