At the global artificial intelligence arena, Baidu's ERNIE-5.0 Preview model has achieved great success. The international competition platform LMArena has released the latest rankings, with Baidu's ERNIE-5.0-Preview-1203 scoring 1451 points and topping the text ranking. Not only does it become the first in China, but it also surpasses many well-known competitors, including Claude-Opus-4-1, GPT-5.2, and GPT-5.1, in creative writing and complex instructions.

The success of ERNIE-5.0 is not accidental. This model was officially launched in November, with a parameter count of up to 2.4 trillion, and uses a unified modeling approach for all modalities. It has the ability to process large text, images, audio, and video inputs. In applications related to creative writing and tasks, ERNIE-5.0 can accurately understand user needs and generate text content that is structured and logically rigorous, performing very well.

image.png

In the historical rankings of LMArena, ERNIE-5.0 had previously tied for second place globally, showing strong competitiveness. In visual understanding, it also performed well, being on par with internationally leading models such as Claude-Sonnet-4 and GPT-5-High. Baidu's new model demonstrates the rapid development and innovation capability of China in the field of artificial intelligence, marking that domestic AI technology is continuously moving toward international frontiers.

Notably, the flexibility and diversity demonstrated by ERNIE-5.0 in applications suggest that it may shine in more industries and fields in the future. With the continuous advancement of AI technology, ERNIE-5.0 not only promotes Baidu's own development but also brings new vitality to the global AI ecosystem.

Baidu's ERNIE-5.0 success is an important breakthrough for China's AI industry, demonstrating the ambition of domestic technology companies in fierce competition.