>>16767478
IQ is a seriously outdated metric.
If you release a new version of ChatGPT, does that constitute "training" or is the new version simply more "intelligent"?
Here's your Einstein IQ btw.

ChatGTP5 still fails the strawberry test because its model can't reach beyond its tokenizer.

It fails wherever computers should be traditionally good at.

I have zero confidence in this piece of artificial idiocy.