← back
x.comQian LiuThu, Jul 2, 2026, 10:30 PM PDT
score 16.1
42likes4RT1reply

EdgeBench benchmark tests AI agents on 12-72 hour tasks

Original: Nice long-horizon benchmark!

Source: x.com

Writing ELI5 summary…

EdgeBench benchmark tests AI agents on 12-72 hour tasks · TinyNews · TinyNews