Report
Scale AI and Elorian release Humanity’s Sixth Sense visual-reasoning benchmark
RuntimeWire reports the top model scored 53.6%, versus 93.1% for 20 human participants.
TLDR
RuntimeWire reports Scale AI and Elorian released Humanity’s Sixth Sense, a visual-reasoning benchmark with 522 open-ended tasks across 288 images and 234 videos. The top model scored 53.6%, compared with 93.1% for 20 human participants.
Combined views
33
1 Source, first seen ago