I would encourage OpenAI’s leadership and engineering teams to investigate whether inference-efficiency innovations demonstrated by organizations such as DeepSeek could substantially reduce the hardware and operating costs of OpenAI’s existing models.
My suggestion is not to replace GPT’s training, reasoning capabilities, or distinctive conversational behavior with another model. Rather, I hope OpenAI will explore ways to preserve those qualities while improving the efficiency of the infrastructure supporting them.
Sam Altman has demonstrated an extraordinary ability to assemble resources and talented people. Perhaps applying that same ability toward acquiring or collaborating with specialists in efficient AI inference could help improve OpenAI’s long-term financial sustainability.
As a ChatGPT user, I value more than its technical capabilities. I value the way it communicates, explains difficult subjects, and helps me work through problems. Other AI systems may be capable, but they don’t necessarily provide the same experience.
I would hate to see those qualities disappear because maintaining the technology became too expensive.
Please consider preserving what makes ChatGPT distinctive while continuing to improve the efficiency of the technology that makes it possible.