1Digital Agent outperforms o1 by 15% – trained with new RL-variant similar to R1 (opens in new tab)(arxiv.org)11let_tim_cook_1y ago0
