Their proposed price increases are insane and yeah even though they are getting lots of funding right now, they arent covering their expenses with subscriptions. I cant imagine they would successfully charge regular users that much without kicking 99% of them off their platform. Now that would be dystopian to me… to price out regular users when their model uses the same computing power.

You are saying they are overstating their models ability? My understanding of the claim is that the model just makes less simple arithmetic mistakes. Ive still noticed hallucinations and mistakes when assisting with my code but to be fair the language im using has limited documentation. I dont see their claims as exaggarated yet but id be lying if i said i have used the new preview model enough to understand it. Its certainly slower…