OpenAI introduces GPT-5.6 models in kiro to enhance developer efficiency

The OpenAI and AWS collaboration has introduced the GPT-5.6 model family within Kiro, a software development agent designed to bring engineering rigor and quality to AI-native coding at scale. The update includes the latest OpenAI flagship models—Sol, Terra, and Luna—integrated into Kiro’s development workflows, which support planning, building, reviewing, and testing software. These models aim to help developers produce higher-quality code with fewer iterations and better value per token. GPT-5.6 is noted for delivering more useful work per token, with improved performance per dollar and on-demand capabilities for complex tasks. Developers can apply these features to long-running projects grounded in specific requirements, codebases, and team standards. Kiro translates high-level intent into clear requirements, technical designs, and executable tasks, providing structured context that helps GPT-5.6 understand the system’s intended functionality and the final implementation goals. Testing on Terminal-Bench 2.1 showed that GPT-5.6 Terra completed tasks in Kiro at approximately 82% lower cost. Kiro’s spec-driven approach ensures models are grounded in clear requirements, technical designs, and task context from the start, leading to faster working solutions with fewer errors. This results in more completed work, less wasted effort, and better value from coding sessions. OpenAI and AWS plan to continue improving the performance of OpenAI models in Kiro to enhance developer value across the software development lifecycle. The GPT-5.6 family is now available in Kiro, with access at https://kiro.dev/.