StoryResearch
ERNIE 5.0 ranks 8th on LMArena's text leaderboard, the only Chinese model in the top 10
The ERNIE-5.0-0110 version scored 1,460 on LMArena's crowd-voted text leaderboard, placing 8th worldwide, ahead of GPT-5.1-High and Gemini 2.5 Pro. It ranked 2nd in maths. The November preview had placed 24th.
- Score 1,460, #8 globally and the only Chinese model in the top 10; #2 in maths.
- Baidu said ERNIE was handling 16.5 billion API calls a day.
- Still behind the newest US models such as GPT-5.2 and Gemini 3.