Select Language:
On September 9th, Baidu announced the official launch of its latest deep thinking model, Wenxin Da Model X1.1. This new iteration is built upon the foundation of the Wenxin Da Model 4.5, with significant enhancements in accuracy, instruction compliance, and intelligent capabilities. The upgrade marks a notable step forward in Baidu’s AI development efforts.
Compared to its predecessor, Wenxin X1, the X1.1 version claims to improve factual accuracy by nearly 35%, increase instruction-following ability by over 12.5%, and boost intelligence functions by close to 10%. These improvements are designed to provide more reliable responses and better adherence to user commands.
Currently, the Wenxin Da Model X1.1 is available to business clients and developers via Baidu’s Qianfan platform on Baidu Smart Cloud, opening the door for widespread commercial and development applications.
Earlier this year, in March, Baidu introduced the Wenxin Da Model 4.5 and Wenxin X1 models—both accessible for free on the Wenxin Yiyan official website. The 4.5 version is notable as Baidu’s first native multimodal large model, with significant capabilities in multimodal understanding, text, and logical reasoning. Its performance even surpasses GPT-4.5 in several tests, with API costs at only about one percent of GPT-4.5’s pricing. Meanwhile, the Wenxin X1 is designed to compete with deep thinking models like DeepSeek-R1, supporting multimodal inputs and multiple tool integrations. Its API services are approximately half the cost of R1, making it an attractive option for businesses seeking advanced AI solutions.
Images accompanying the announcement show a visual of Baidu’s new model, highlighting improvements in fact accuracy and instruction adherence, reflecting the strides Baidu is making in the expanding field of AI-driven language models.
Add us on ChatGPT
Add us on Perplexity
