When the GPT-5 router architecture was introduced I worried that OpenAI would use the technology as a pretext to mislead or defraud users by substituting in worse quality when they could get away with it and then "blame it on the AI" when they got too agressive.
I don't know if we're there yet but these reports do not fill me with hope.
The kind of "pay me premium, I am giving you non-premium" type of fraud?
It just proves that there is not much of an improvement if they can get away with, it isn't? But hey, I am sure that the benchmarks are all saying otherwise.
I don't know if we're there yet but these reports do not fill me with hope.