The Cheapest AI Models Reveal a Critical Vulnerability

August 6, 2025

This blog post is the work of an authentic dinobaby. Sorry. Not even smart software can help this reptilian thinker.

I read “Price Per Token,” a recent cost comparison for smart software processes. The compilation of data is interesting. The two lowest cost services using a dead simple method of Input Cost + Output Cost averaged Open AI GPT 4.1 nano and Gemini 2.- Flash. To see how the “Price Per Token” data compare, I used “LLM Pricing Calculator.” The cheapest services were OpenAI – GPT-4.1-nano and Google – Gemini 2.0 Flash.

I found the result predictable and illustrative of the “buying market share with low prices” approach to smart software. Google has signaled its desire to spend billions to deliver “Google quality” smart software.

OpenAI also intends to get and keep market share in the smart software arena. That company is not just writing checks to create a new class of hardware for persistent AI, but the firm is doing deals, including one with Google’s cloud operation.

Several observations:

Google and OpenAI have real and professional capital on the table in the AI Casino
Google and OpenAI are implementing similar tactics; namely, the good old cut prices in the hope of winning market share while putting the others in the game to go broke
Google and OpenAI are likely to orbit one another until one AI black hole absorbs or annihilates the other.

What’s interesting is that neither firm has smart software delivering rock solid results without hallucination, massive costs, and management that allows or is helpless to prevent Meta from eroding both firms by hiring key staff.

Is there a fix for either firm? Nope and Meta’s hiring tactic may be delivering near fatal wounds to both Google and OpenAI. Twins can share similar genetic weaknesses. Meta may have found one —- paying lots for key staff from each firm — and is quite happy implementing it.

Stephen E Arnold, August 6, 2025

Written by Stephen E. Arnold · Filed Under AI, Financial, News

Comments

Got something to say?

Search the site
Subscribe to Beyond Search
Feature archive
News archive

Stephen E. Arnold monitors search, content processing, text mining and related topics from his high-tech nerve center in rural Kentucky. He tries to winnow the goose feathers from the giblets. He works with colleagues worldwide to make this Web log useful to those who want to go "beyond search". Contact him at sa [at] arnoldit.com. His Web site with additional information about search is arnoldit.com.