Comment by Simplita

6 hours ago

Big models keep getting better at benchmarks, but reliability under messy real world inputs still feels stuck in place.

0 comments