What does a human knowledge worker actually cost per token?
I ran the math, and the answer stopped me cold: $8,333 per million tokens.
Here’s how that breaks down. The average knowledge worker processes about 36,000 words per day—emails, meetings, documents, Slack messages. Convert that to tokens (the unit AI models use), and you get roughly 48,000 tokens daily.
At a $100,000 salary across 250 working days, that’s $400 per day of human cognition. Divide by those 48,000 tokens, scale it up, and you land at $8,333 per million tokens.
GPT-4o? $10 per million tokens.
That’s an 833x cost difference.
The $20 Knowledge Worker
But here’s where it gets even more interesting for individuals.
You don’t need to pay per token at all. ChatGPT Plus, Claude Pro, and Gemini Advanced all cost $20 per month. For that price, you get 40-80 hours per week of AI runtime—essentially unlimited for most personal use cases.
That’s a full-time knowledge worker for the cost of a few coffees.
No hourly rate. No per-task billing. Just $20 and you have a research assistant, writing partner, analyst, and brainstorming buddy available whenever you need one.
It’s Only Getting Cheaper
If you think this sounds too good to be true, consider the trajectory.
Source: Andreessen Horowitz - LLMflation
In November 2021, GPT-3 cost $60 per million tokens. By October 2024, you could get equivalent performance for $0.06 per million tokens. That’s a 1,000x cost reduction in three years.
When GPT-4 launched in March 2023, it cost $30-60 per million tokens. Sixteen months later, GPT-4o delivered the same capability for $3-10 per million tokens—a 90% drop.
The pattern is clear: AI gets roughly 10x cheaper every year. And there’s no sign of that slowing down.
Stop Counting Tokens
I see people agonizing over token usage, trying to write shorter prompts, worried about “wasting” AI capacity.
Stop.
At current prices—especially with subscription plans—the cost of using AI is effectively zero for individuals. The real cost is not using it. Every hour you spend on tasks that AI could handle is an hour of $8,333-per-million-token cognition you’re burning on commodity work.
The economics have fundamentally shifted. Intelligence is no longer scarce or expensive. Learning to leverage it effectively—that’s where the value is now.