On December 23, LMArena Large Model Arena unveiled its latest rankings, revealing that the new iteration of the ERNIE large model, ERNIE-5.0-Preview-1203, has claimed the top position in the Chinese text rankings with an impressive score of 1451. This achievement also places it ahead of leading domestic and international models, including Claude-Opus-4-1 and GPT-5.2, in areas such as creative writing and the execution of complex instructions. Launched in November of this year, the model is equipped with a staggering 2.4 trillion parameters and offers comprehensive support for full-modal input and output across text, images, audio, and video formats. Prior to this, it had shared the second place globally and held the top spot domestically in the text rankings, while also securing the first position in the domestic visual understanding rankings. Sources close to the matter have indicated that the official version of the ERNIE Large Model 5.0 is set to be released in January next year.