在Agent编程评估Terminal-Bench 2.0中取得了最高分，并在“人类最后考试”中领先所有其他前沿模型。 在MRCR v2 8-needle 1M基准测试——大海捞针——中，Opus 4.6得分76%，而Claude Sonnet 4.5只有18.5%。

Anthropic’s Cowork AI assistant sent shockwaves through Wall Street this week over concerns it could replace specialized software packages, such as for legal or financial analysis. Now Anthropic is ...

In a significant move for the software industry, AI company Anthropic (ANTH.PVT) has unveiled its latest model, Opus 4.6, which it touts as the "most capable model for all enterprise and knowledge ...