AI Models Outperform Frontier Models in Software Testing | Dark Hacker News