What Went Wrong with OpenAI's GPT-3 Pricing Model

A Missed Opportunity

OpenAI recently released details about their Pricing Model:

Although I’m excited about these transparent pricing details, as well as OpenAI’s generous policy which allows users to retain copyright over GPT-3 outputs, this still is not as low as I would have liked it.

I would have liked billing to be a fraction of what it is now ($10 a month for 2M credits) because I think it would have allowed GPT-3 to be at the center of many commercial applications, emerging startups, side projects, as well as several creative disciplines alike. It could have been a quick API you just toss in along with your application, no matter what you’re working on. At $10, I believe it would have been so essential, almost like paying for the electricity for your home or your current cell phone plan. Table stakes.

Personally, I can just see GPT-3 being used everywhere. Any discipline, any daily routine, or any level of technical experience. Even for small script utilities or virtual day-to-day assistant type work, GPT-3 is just a handy resource to have around. Even to ask real life personal questions before bed, which is something I actually do, it’s a handy resource and worthwhile to have around.

At $10, it is an obvious, “yes”. I think many, possibly millions of people, would have no issue subscribing to it. However, at $100, there is greater thought and consideration the mass market has to overcome in order to come around. I also think $10 would attract more ordinary types of people, outside of the tech industry, whereas $100 puts more weight on the pricing decision which might scare ordinary non-tech consumers away.

100 dollars a month means fewer subscribers, fewer interesting use cases being discovered, and an overall weaker global technological dependency for something I believe which will be so critical in the future.

$10 is an easy choice, $100 is a difficult one. $10 is a quick hold over the entire market, $100 is only a subset of all the potential users out there. Finally, $10 would have priced GPT-3 more like an essential monopolistic utility, whereas $100 prices it like just any other SaaS based company. Like Basecamp, or GSuite, or something.

But maybe I’m just an optimist and a true believer, which I’m willing to accept is the likely case. This also means I’ll still be paying the $100 a month for GPT-3 access, so, maybe OpenAI does actually know what they’re doing and how valuable their GPT-3 model is afterall.