It is getting to a point where the AI models have little advantage over each other in the benchmarks. While some suggest that a ceiling is being reached, I see it more as approaching a new phase, to continue to improve.

I wonder what it is?