Aug. 7 (Bloomberg) -- The Wall Street Journal reported today that people with knowledge of the situationBaiduThe plan is to launch by the end of this month (end of August 2025) a newinference modelThe model is capable of handling more complex tasks to gain a foothold in the competition with the likes of DeepSeek and OpenAI.

The report added that Baidu also plans to roll out an updated version of its core base model in the coming months, theThe name is Ernie 5.0.(Note: Ernie is Wenxin).
Baidu released the "strongest ever" Wenshin Big Model 4.5 and Wenshin Big Model X1, which supports deep thinking, in March this year:
- Wenxin Big Model 4.5 is Baidu's first native multimodal big model, with significant improvements in multimodal comprehension, text and logical reasoning, etc. It outperforms GPT4.5 in a number of tests, and the price of API calls is only 1% of GPT4.5;
- Wenxin Big Model X1 is a deep thinking model that benchmarks DeepSeek-R1 in terms of performance, and also supports multi-modal, multi-tool call capability, with API calls at about half the price of R1.