跳到正文
原文
ArtificialAnlys· @ArtificialAnlys · X·· 5 小时前AI 评分44

HealthBench评测:GPT-6 Sol沟通分高但完整性最低

AI 导读

在 HealthBench 上,GPT-6 Sol (max) 写出的答案最短(约 1,000 字符),原始沟通质量得分最高(81%),但完整性得分最低(34%)。在 14 个模型中,有 13 个模型得分最低的主题是 Context Seeking 或 Global Health。

正文 · 原文

On HealthBench, GPT-6 Sol (max) writes the shortest answers (~1,000 characters) and has the highest raw Communication Quality score (81%), but the lowest Completeness score (34%). Context Seeking or Global Health is the lowest-scoring theme for 13 of the 14 models

来源:ArtificialAnlys · x.com