On July 30, 2026, OpenAI cut the API price of two of its three GPT-5.6 tiers. Its developer changelog states that starting July 30, GPT-5.6 Luna costs 80 percent less and GPT-5.6 Terra costs 20 percent less. The change applies to both the v1/responses and v1/chat/completions endpoints. The flagship Sol tier was not repriced.
The same changelog entry introduces Fast mode, which replaces the previous Priority Processing option. Fast mode offers up to 2.5 times faster speeds than standard processing at twice the price, and requests already marked as priority migrate to it automatically, so no integration breaks.
The timing is the story. GPT-5.6 launched on July 9, 2026, so the cheapest tier lost 80 percent of its price roughly three weeks after general availability. Frontier-lab pricing has historically fallen in steps of six to twelve months; a three-week cut on a just-launched generation is a different cadence, and it lands in the same fortnight as a wave of capable open-weight releases from Chinese labs.
For anyone budgeting AI spend, the practical lesson is to stop treating list prices as fixed inputs to a business case. Workloads that did not clear an ROI threshold at 1 dollar per million input tokens clear it at 20 cents, and the interval over which that reprice can happen is now measured in weeks. Contracts, forecasts, and vendor-selection decisions built on a six-month price assumption need shorter review cycles.