It is a tiered price list; pricing power remains, but the top tier has started to give. On 7/30 OpenAI cut Luna 80% and Terra 20% and left Sol unchanged; VentureBeat reported that on 9/22 GPT-6 Sol's list price was halved, so 'the top does not cut' is now being tested by its own falsification clause.
The mainstream looks at the size of the cut and calls it a price war; the diary looks at the distribution. OpenAI attributed the cut to efficiency, yet efficiency is shared across the family and Sol did not move; a 20% serving-cost drop plus 15% token efficiency does not reach 80%. The author judges the gap to be strategic pricing, capital buying share. That is the author's judgment, not an official statement.
OpenAI 7/30 (CNBC, OpenAI official): GPT-5.6 Luna down 80%, Terra down 20%, Sol unchanged with a Fast mode at 2x the price. VentureBeat 2026-09-22, reported figures: GPT-6 Sol at $2/$10 versus the predecessor's $4/$20, which OpenAI calls a long-term list price, not an introductory one; ENTRY 118 recorded the predecessor at $5/$30, and interim repricing was not tracked by this site.
For founders: do not bet gross margin on the mid-tier spread; either move value up to top-tier capability or take the volume at the bottom. For investors: watch the distribution, not the headline percentage, and whether the top tier follows; on reported figures Sol has been halved, so count model companies' multi-year compute commitments as quasi-debt inside EV.