GPT-5.4 matches human experts 82% of the time, up from 38% | Tech Twitter
GPT-5.4 matches human experts 82% of the time, up from 38%
Press Space to continue
To me, the question of whether Microsoft will give access to the most recent/best models is big: Given that GPT-5 beat or tied humans in expert tasks (GDPval) less than 38% of the time while months later, GPT-5.4 beat or tied human experts 82% of the time, this really matters.