The outcome of a poll on this topic would seem to me to likely have a strong correlation to the poll demographics. I assume that I am atypical of the general manifold demographic so I am curious if manifold expectations were higher or lower for GPT-5 than expectations, if any, comming from the general population, which i expect to be less informed on such topics than the manifold demographic.

But i could he off even in that assumption. Maybe a manifold poll could be representative of the general pop, where llms are concerned.

@ShitakiIntaki Only a very small subset of the general population had expectations for GPT-5 and I expect that subset's demographics to be much closer to the population of Manifold users than to the general population.

Yep

Definitely no, but didn't really fall short either

I am now using GPT-5 Thinking as my default model. I was not expecting that to happen.

It's alright but the expectations were pretty high to begin with.

i'm not coming at this from an advanced angle or anything but gpt-5 no thinking is did not exceed expectations, and part of this is how sam framed it 'you will use this for most purposes'. it continually fails what I want it to do.

Thinking however rarely ever misunderstands me, and hardly takes any more time at all! Thinking is very impressive imo. Better than nearly any model ive just toyed around with. Did it exceed expectations... maybe slightly?

On the whole that makes me say 'no'

actually it's worse than 4o

@realDonaldTrump disagree! What are you trying to do with it?