As recorded at collection: OpenAI cuts GPT-5.6 Luna prices by 80%

Luna, the fastest and least expensive model in the GPT-5.6 family, will have its price reduced by 80%, while Terra, the balanced model for everyday work, will be 20% cheaper. The same reductions will be reflected in usage calculations for Codex and ChatGPT Work, so using the two models for the same task will deduct fewer credits from your subscription limit.

Starting July 30, Terra will cost 2 dollars per 1 million input tokens and 12 dollars per 1 million output tokens. Luna will cost 0.20 dollars per 1 million input tokens and 1.20 dollars per 1 million output tokens, while Sol prices will remain unchanged. The reduced rates will apply when you use the API as usual, with no separate application required.

The API is introducing a fast mode to replace the existing priority processing service. For Sol, it is 2.5 times faster than standard processing and costs twice as much, with the same answer quality. Requests previously sent with the priority processing flag will automatically use fast mode, so no changes are needed. ChatGPT and Codex subscription prices and usage limits are unchanged. AWS price changes will be rolled out in stages starting later today.