Amazon Web Services has announced the general availability of OpenAI GPT-5.6 models—Sol, Terra, and Luna—on Amazon Bedrock. This release includes explicit prompt caching, which allows users to control which parts of their prompts are cached and reused.
Key Points
- OpenAI GPT-5.6 Sol, Terra, and Luna models are now generally available on Amazon Bedrock.
- The release introduces explicit prompt caching.
- Explicit prompt caching provides precise control over which prompt parts are cached.
- The feature enables reuse of cached prompt segments.
- Explicit caching can help reduce inference costs.
Context
According to the AWS Machine Learning Blog, the new caching mechanism is designed to give users granular control. This allows for specific sections of a prompt to be stored and recalled, rather than caching entire prompts indiscriminately.
Why It Matters
This update offers builders a method to potentially optimize resource usage and manage operational expenses by reducing the computational load associated with repeated prompt elements.
What To Do
- Learn how to get started with the newly available GPT-5.6 models on Amazon Bedrock.
- Set up explicit prompt caching for new applications.
- Explore migrating existing GPT workloads to utilize the new caching feature.
- Evaluate the impact of explicit caching on inference costs for current projects.