Bookmark · 13 November 2024, 10:31 New secret math benchmark stumps AI models and PhDs alike - Ars Technica ml-gwb-predictions