No money, no honey
I’ve been seeing a slew of complaint posts from ChatGPT users (though most seem to be on the free tier) about usage limits getting absurdly tight.
Then I checked out the new Gemini iOS app a few days ago and, lo and behold, Gemini has basically become the new Claude when it comes to limits. Aside from the rolling 5-hour limit, there’s now a weekly one too. On Gemini!! The one bot I always went to for near unlimited use.
Same issue with Grok. And Claude has always been Claude.
So it seems the squeeze on token costs is happening across all mainstream chatbots.
Which got me thinking: is this the end of “relational AI” use? People on this sub complain incessantly about model changes, but let’s say hypothetically some company could recreate something like the OG 4o, or *GASP* even something far better, except now the token usage costs an arm and a leg.
Was “relational AI” only possible during the subsidized growth phase of chatbots?
Can this kind of use even survive once companies fully price in its real inference costs?
And no, I don’t believe Chinese AI will stay relatively cheap forever. Nor do I think going local is feasible for the majority of regular users.