Fewer than 7% of paying ChatGPT users regularly picked o3, the strong reasoning modelA model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.Full definition →, largely because "4o" looks like the bigger number. After GPT-5's auto-router shipped, reasoner use hit 24% of paid users and went from near zero to 7% of free users.
"GPT-5" is a router plus a family of models: it decides whether your prompt gets a cheap small model or a slow reasoner. Early on it misfired, so two people asking the same thing got very different answers. If you pay, just pick GPT-5 Thinking yourself.
GPT-4 cost roughly $50 per million tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → at launch. GPT-5 nano costs about 14 cents per million and is more capable. That collapse in marginal cost is what makes free tiers and ad-supported AI arithmetically possible, not any democratization push.
A modern LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → prompt uses about 0.0003 kWh, the same as 8-10 seconds of Netflix, and Google reports per-prompt energy efficiency improved 33x in a year. Water use is far less settled: credible estimates span 0.25mL to 5mL per prompt depending on what you count.
Mollick's recent experiments found the old prompting tricks no longer help. chain-of-thoughtHaving a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.Full definition → scaffolding, politeness, and threats all wash out on frontier models, which now infer what you actually want well enough that elaborate prompting buys nothing on average.
Terms in this piece · Glossary
reasoning model — A model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
chain-of-thought — Having a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.
Why it matters
A clear-eyed framing of what happens when advanced AI becomes as cheap and ubiquitous as web search, with concrete cost/energy figures and capability demos that help builders reason about pricing, adoption, and the shifting baseline of user expectations.
Key quotes
“Within a few days of launch, the percentage of paying customers who had used a Reasoner went from 7% to 24% and the number of free customers using the most powerful models went from almost zero to 7%.”
“When GPT-4 came out it was around $50 to work with a million tokens (a token is roughly a word), now it costs around 14 cents per million tokens to use GPT-5 nano, a much more capable model than the original GPT-4.”
“This is how a billion people suddenly get access to powerful AIs: not through some grand democratization initiative, but because the economics finally make it possible.”
“Powerful AI is cheap enough to give away, easy enough that you don't need a manual, and capable enough to outperform humans at a range of intellectual tasks.”
