OpenAI has announced significant price reductions for two of its GPT-5.6 models, Luna and Terra, as part of an ongoing effort to enhance model efficiency. The company stated that the API price for GPT-5.6 Luna has been cut by 80%, while GPT-5.6 Terra's price has been reduced by 20%.
Under the updated pricing structure, GPT-5.6 Luna now costs $0.20 per million input tokens and $1.20 per million output tokens. This represents a substantial decrease from its previous rates of $1 per million input tokens and $6 per million output tokens. Similarly, GPT-5.6 Terra's price has dropped from $2.50 to $2 per million input tokens and from $15 to $12 per million output tokens.
These price adjustments also impact how usage is calculated for customers utilizing Codex and ChatGPT Work. OpenAI indicated that new tasks processed by these models will consume less of a customer's allowance, enabling more work to be completed within the same quota.
Furthermore, OpenAI is upgrading the Auto-review feature within the ChatGPT application and the Codex command-line interface (CLI) from GPT-5.4 to GPT-5.6 Luna. This upgrade is projected to reduce associated costs by approximately tenfold.
In addition to the price cuts, OpenAI has introduced a "Fast mode" for API customers using GPT-5.6 Sol. This new mode offers up to 2.5 times faster processing compared to standard Sol processing, without compromising the model's intelligence. However, there are no changes to the standard pricing for GPT-5.6 Sol. The enhanced performance of Sol Fast mode comes at twice the standard API price, making it particularly suited for time-sensitive applications such as coding, research, and agentic workloads. For other use cases, the standard GPT-5.6 Sol is deemed sufficient.
OpenAI attributes the efficiency gains that enabled the Luna and Terra price reductions to recent improvements made to GPT-5.6 Sol. The company's internal testing also places Luna at the top of its intelligence index among comparable models, despite its significantly lower cost per task.
