DayNews.aiAI agents tested over hours, not steps
A new benchmark tests AI agents on complex, multi-hour tasks and rewards partial progress, not just final outcomes.
DayNews.aiDayNews.aiDayNews.aiA new benchmark tests AI agents on complex, multi-hour tasks and rewards partial progress, not just final outcomes.