ERNIE Bot Leads Tsinghua University’s LLM Report Rankings in China

  • Baidu’s ERNIE Bot 4.0 claims the top position in Tsinghua University’s evaluation of large language models (LLMs) in China.
  • Despite domestic success, ERNIE Bot faces stiff competition from international models like OpenAI’s GPT-4 and Anthropic’s Claude-3.
  • The evaluation, utilizing the SuperBench framework, highlights ERNIE Bot’s strengths in Chinese language comprehension but reveals shortcomings in semantic understanding and coding abilities.
  • ERNIE Bot excels in specific capabilities within the domestic landscape, notably in Chinese language comprehension and adherence to human commands.
  • Baidu celebrates ERNIE’s widespread adoption with over 200 million users since its launch, indicating a narrowing gap between Chinese and international LLMs.

Main AI News:

Baidu’s ERNIE Bot 4.0 continues to make waves in the field of large language models (LLMs), securing the top position among Chinese counterparts in Tsinghua University’s recent evaluation. However, as the international stage witnesses fierce competition from models like OpenAI’s GPT-4 and Anthropic’s Claude-3, ERNIE Bot faces challenges in asserting its global dominance.

The evaluation, conducted by Tsinghua University’s Basic Model Research Centre in collaboration with the Zhongguancun Laboratory, utilized the SuperBench framework to assess 14 prominent LLMs. While ERNIE Bot demonstrates strengths in certain areas, such as Chinese language comprehension, it trails behind international peers in semantic understanding, coding proficiencies, and responsiveness to human instructions.

ERNIE Bot’s supremacy extends to specific capabilities within the domestic landscape. For instance, it leads the pack in Chinese language comprehension, boasting a significant lead over competitors like Zhipu AI’s GLM-4. In mathematical prowess and semantic comprehension, ERNIE 4.0 shares the top spot with Claude-3, outperforming GPT-4 models. Noteworthy is ERNIE 4’s commendable performance in adhering to human commands, securing the second position closely behind GPT-4.

Furthermore, ERNIE 4 distinguishes itself with exceptional safety and security capabilities, outpacing competitors with a significant lead.

Baidu celebrates ERNIE’s widespread adoption, with over 200 million users leveraging its capabilities since its launch. While disparities persist between Chinese and international LLMs, the Tsinghua study suggests a narrowing gap. With ERNIE Bot leading the charge alongside other prominent contenders, such as Alibaba’s Tongyi Qianwen 2.1 and Moonshot AI’s Kimi chatbot, the landscape of AI in China and globally promises continued evolution and competition.


The dominance of Baidu’s ERNIE Bot in Tsinghua University’s LLM report rankings signals its strong presence in the Chinese market. However, the stiff competition from international models underscores the need for continuous innovation and improvement to maintain global competitiveness. Baidu’s success with ERNIE Bot reflects the evolving landscape of AI in China, with promising implications for the market’s future development and competitiveness.