Recently, OpenAI has formed a deep collaboration with the software development agent platform Kiro, announcing the full integration of the new GPT-5.6 model family into Kiro. This update brings the latest flagship model series, including Sol, Terra, and Luna, seamlessly embedded into the complete development workflow for team planning, building, reviewing, and testing software.

As a core upgrade, GPT-5.6 achieves a significant increase in token output efficiency, offering stronger performance and higher cost-effectiveness while providing the ability to handle complex tasks on demand. On the Kiro platform, developers can apply this capability to long-running development tasks and deeply integrate it with their project requirements, codebase, and team standards. Kiro is able to transform high-level design intentions into clear operational requirements, technical designs, and executable tasks, allowing GPT-5.6 to accurately understand the team's construction goals and final implementation path.

With the powerful empowerment of GPT-5.6 in Kiro, developers can efficiently complete multiple core tasks. These include directly transforming product ideas into structured implementation plans, handling complex multi-step coding tasks with high continuity, injecting order into AI programming through a specification-driven approach, fully utilizing the codebase and team standard context, reviewing and fine-tuning the model's work results at key checkpoints, and using attribute-based testing to verify the correctness of the implementation.

To achieve the best experience, OpenAI has closely collaborated with Amazon Web Services (AWS) to perform low-level optimizations on the Kiro environment and OpenAI models. Public test data shows that in the Terminal-Bench 2.1 test, GPT-5.6 Terra successfully completed all tasks in Kiro, with costs reduced by approximately 82%. This specification-driven approach anchors the model from the start on clear requirements and context, helping developers achieve practical solutions with fewer operational errors and faster results.

Swami Sivasubramanian, Vice President of AWS Agent AI, stated that the team has always been committed to introducing the latest foundational models for developers to accelerate AI-native development processes. Colleen Kapase, Vice President of Global Strategic Partnerships and Ecosystem at OpenAI, pointed out that integrating GPT-5.6 into Kiro allows developers to more flexibly match the model's intelligence level, speed, and cost to each stage of software development, helping technology teams achieve higher returns when investing in AI. In the future, both parties will continue to deepen their collaboration, further unlocking the huge value of artificial intelligence throughout the entire software development lifecycle.