Post
The New Stack
thenewstack.io
did:plc:kfaqksbyr3zwzvzj3wpbpvac
Coding agents shine on demos but fall apart on long tasks. A look at the endurance gap, Berkeley's strict benchmark, and the race to fix it.
https://bit.ly/4oH3RlJ
2026-06-19T22:00:31.973Z