AI 导读
Opus 5.5 每任务使用 15.6M tokens,而 Opus 5 为 11.4M(+37%)。缓存输入从 10.9M 升至 14.6M,输出则翻倍以上,从 137k 增至 333k。https://t.co/LMWbmpCx4B
正文
Opus 5.5 uses 15.6M tokens per task vs 11.4M for Opus 5 (+37%). Cached input rises from 10.9M to 14.6M, while output more than doubles, from 137k to 333k. https://t.co/LMWbmpCx4B
来源:@ArtificialAnlys · x.com