Baidu, a Chinese multinational technology company specializing in Internet services and Artificial Intelligence, has unveiled two new AI models as it intensifies efforts to strengthen its position in China’s highly competitive AI landscape.
The two new models are Ernie 4.5, the latest version of the company’s foundational model first released two years ago, and a new reasoning model, Ernie X1.
The roll-out of these models, marks a shift in Baidu’s strategy, with the company embracing an open-source approach to keep pace with emerging players.
Register for Tekedia Mini-MBA edition 17 (June 9 – Sept 6, 2025) today for early bird discounts. Do annual for access to Blucera.com.
Tekedia AI in Business Masterclass opens registrations.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register to become a better CEO or Director with Tekedia CEO & Director Program.
Baidu claims that Ernie X1’s performance is “on par with DeepSeek R1 at only half the price,” and it touts Ernie 4.5’s “high EQ,” allowing the model to understand memes and satire. Both models have multimodal capabilities, allowing them to process video, images, and audio, as well as text.
Recall that Ernie Bot garnered one million downloads within 19 hours of its release, and Baidu’s shares rallied by over 4% on the day of release. This positive response signified the potential of Baidu’s AI initiatives, which were expected to drive more traffic to its search engine and boost ad revenues, however, the momentum wasn’t sustained as it struggled to keep up with the intense competitive AI space.
Industry analysts see the launch of Baidu’s two new AI models as a necessary move for the company, which has struggled to maintain dominance and widespread adoption, since the launch of its AI chatbot Ernie in 2023. Once positioned as China’s answer to OpenAI’s ChatGPT, Ernie has since been outpaced by offerings from innovative startups like ByteDance (TikTok parent company), Alibaba, and Deep Seek, amongst others.
“Baidu has been slow to adapt to market shifts, allowing rivals to gain ground,” said Wei Sun, an Al analyst at Counterpoint Research. “The success of its new models will depend on whether they truly deliver better performance and cost efficiency.”
One of Baidu’s key challenges has been its reliance on proprietary Al development, a contrast to competitors who have embraced open-source methodologies. Unlike closed-source models that are developed in isolation, open-source Al benefits from community contributions, accelerating improvements and adoption.
“Baidu initially resisted the open-source movement, believing its proprietary approach would offer a competitive edge,” said Kai Wang, a senior equity analyst at Morningstar. “However, companies like DeepSeek have demonstrated that open models can be just as powerful, if not more so while being significantly cheaper to develop.”
Regulatory pressures have also slowed Baidu’s Al advancements. The company has had to navigate government policies while securing funding in a fast-evolving market, diverting resources away from innovation.
In response to these setbacks, Baidu recently announced plans to open-source its next-generation Ernie model starting June 30.
This shift aligns it with major Chinese tech firms like Alibaba, Tencent, and DeepSeek, all of which have already embraced open Al development. “Baidu is now following the path of its competitors,” said Lian Jye Su, chief analyst at Omdia. “It remains to be seen whether this pivot will be enough to restore its standing in China’s Al race.”
As the competition between artificial intelligence (AI) chatbots intensifies, several companies and researchers in China are making progress on building Chinese-language AI models.
With growing competition from domestic and international AI firms, Baidu aims to solidify its position through continuous innovation. This latest launch of two new AI models could help it regain relevance provided its new models live up to expectations.