跳到正文
RCreddit.com·
暂不在当前实时榜单

Did Anthropic just find a new way to benchmax Opus 5.5 without actually giving subscribers that performance?

AI 摘要

A Reddit user questioned whether Anthropic subscribers actually receive the performance benchmarked for Opus 5.5. The concern arises because Anthropic benchmarks Opus on "Max" settings, but most subscribers use "Medium/High" due to "Max" consuming 6x or more usage and quickly depleting their limits. This discrepancy leads to speculation about whether users are truly experiencing the model's advertised capabilities.

为什么是这条

This report uniquely highlights a potential discrepancy between Anthropic's internal Opus 5.5 benchmarks and the actual performance subscribers experience, unlike typical performance reviews.

时间与来源

时间显示为 UTC

显示时区:UTC

本地时区尚不可用,暂时显示 UTC。

收录当时偏移:UTC+02026年9月22日 18:01 UTC

收录
2026年9月22日 18:01
来源类型
开发者社区

本站未收录正文。

前往源站阅读 →
来源·reddit.com