阶跃星辰语音模型登顶全球第一

Core Insights - The article highlights that the model Step-Audio-R1.1 from Jieyue Xingchen has topped the Artificial Analysis Speech Reasoning leaderboard, which evaluates Native Audio Models [1] Group 1: Model Evaluation - The leaderboard serves as a third-party benchmark for assessing Native Audio Models, focusing on their ability to process audio and perform complex logical reasoning [1] - Key evaluation dimensions include accuracy and first packet latency, which are critical for determining model performance [1]