Anthropic Sonnet 5.5 效率争议
精选理由
Anthropic 的 Sonnet 5.5 效率说法与第三方测试结果不符，值得了解详情。
Anthropic 声称 Sonnet 5.5 比前代 Sonnet 5 使用更少 token，因此任务成本更低。但第三方测试机构 Artificial Analysis 显示 Sonnet 5.5 是目前 token 效率最低的模型之一。这一矛盾引发了用户对 Anthropic 声称的质疑。
原文 · Matt Wolfe
I'm very confused by Anthropic's claims... They claim Sonnet 5.5 uses much less tokens and therefore has a lower cost per task than Sonnet 5. However, Artificial Analysis has it basically being the most token-inefficient model out right now. What am I missing? https://t.co/KDdyYShfQu
