Why comparing AI models by price per token is misleading
By
janilowski
Summary
This article argues that comparing AI models by their per-token pricing ($X per 1M tokens) is misleading because each frontier AI lab uses its own tokenizer, meaning the same text input can be tokenized into different numbers of tokens across models. The author warns that this practice can drive up AI API costs and advises companies to look beyond simple per-token price comparisons when evaluating model costs.
Source
Hacker NewsWhy comparing AI models by price per token is misleadingjanilowski.plKey quotes
· 3 pulled$X per 1M tokens is incomparable
Each frontier lab has its own tokenizer, which determines how many tokens a body of text is split into.
It stops being all about the vibes when the API bill hits you.
You might also wanna read
How companies are burning through AI tokens and facing soaring costs
Companies are rapidly burning through AI tokens—units that measure AI model usage—leading to soaring costs as they integrate generative AI i
financialpost.com·28d ago
AI industry faces first price war as users demand value alongside capability
The AI industry is entering its first price war as users prioritize value alongside capability. Frontier US labs like OpenAI and Anthropic l
semafor.com·25d ago
Tokenmaxxing drives up AI costs as enterprises confuse token usage with productivity
The article discusses "tokenmaxxing" — a trend where enterprises treat high AI token usage as a proxy for productivity, leading to inflated
thenewstack.io·1mo ago
Tokenmaxxing drives up AI costs as enterprises confuse token usage with productivity
The article discusses "tokenmaxxing" — a trend where enterprises treat high AI token usage as a proxy for productivity, leading to inflated
thenewstack.io·1mo ago
Tokenmaxxing fails as AI ROI metric; workflow redesign needed for productivity gains
The article argues that 'tokenmaxxing'—using token usage as a proxy for employee AI innovation—has failed to deliver meaningful ROI for comp
fortune.com·1mo agoTokenmaxxing fails as AI ROI metric; workflow redesign needed for productivity gains
The article argues that 'tokenmaxxing'—using token usage as a proxy for employee AI innovation—has failed to deliver meaningful ROI for comp
FORTUNE·1mo ago
Comments
Sign in to join the conversation.
No comments yet. Be the first.