GPT-5.6 Luna is the cost-efficient member of the GPT-5.6 family, positioned between the flagship Sol and the more balanced Terra. It is intended to make the family’s capabilities practical for routine knowledge work while still supporting demanding tasks through its reasoning-oriented design. Luna Pro uses the same underlying model with a higher-quality reasoning mode for complex requests, offering a clear path to stronger responses when greater reasoning effort is warranted.

The practical appeal is a combination of economical operation, broad input support, and the ability to handle substantial working context. It is a good fit for teams that want a capable general-purpose model for everyday work without moving to the top-tier setting, while still having the option to select a more intensive reasoning configuration for harder problems. The available evidence does not establish specific training scale, parameter counts, or benchmark results for Luna itself.