AI agents ask better questions than KPMG accountants in audit trial 10/18 • europeanbusinessmagazine.com
DaisyRose-27B: 1,200-task RL took Qwen3.8 from 5.6% to 12% on Terminal-Bench 4.1 10/18 • www.radixark.com