OpenAI's GPT-5.6 model family, including Sol, Terra, and Luna variants, is now integrated into AWS's Kiro coding agent. The update improves price-performance for developers using Kiro's spec-driven workflow to plan, build, review, and test software. Benchmarks on Terminal-Bench 2.1 show GPT-5.6 Terra completing successful tasks in Kiro with roughly 82% cost reduction, the result of joint optimization work between OpenAI and AWS.
Questions this post answers
What cost reduction does GPT-5.6 Terra achieve in Kiro compared to before?
GPT-5.6 Terra completes successful tasks in Kiro at roughly 82% cost reduction, based on Terminal-Bench 2.1 testing. This improvement comes from joint optimization work between OpenAI and AWS, combining Kiro's spec-driven approach with the GPT-5.6 model family's better price-performance per token. Developers evaluating AI coding agent costs can track model pricing shifts like this on daily.dev.
Which models make up the GPT-5.6 family available in Kiro?
The GPT-5.6 family available in Kiro includes three models named Sol, Terra, and Luna. These are OpenAI's latest flagship model series, brought into Kiro's development workflows to help teams plan, build, review, and test software with stronger performance per dollar than prior models. Teams comparing AI coding agent model lineups can follow updates like this on daily.dev.