China’s Baidu (NASDAQ:BIDU) is set to release an upgraded version of its artificial intelligence model, Ernie 4.5, in mid-March, featuring enhanced reasoning and multimodal capabilities. According to a source familiar with the matter, this iteration will improve the model’s ability to process text, video, images, and audio.
Baidu recently announced plans to gradually roll out Ernie 4.5 in the coming months, with an official open-source launch slated for June 30. This shift marks a strategic move as the company faces increasing competition, particularly from Chinese AI startup DeepSeek, which claims its models rival top U.S. AI systems at a fraction of the cost.
Despite Baidu’s early lead in China’s AI chatbot market, adoption has been slow, even though the company asserts Ernie’s performance is comparable to OpenAI’s GPT-4. The emergence of DeepSeek has intensified market competition, prompting Baidu to rethink its AI strategy. CEO Robin Li, who previously advocated for a closed-source model, has now embraced an open-source approach following DeepSeek’s rapid success.
Baidu is not alone in this pivot. Alibaba (NYSE:BABA) also announced this week that it will open-source its latest AI model, Wan 2.1, which specializes in generating video and images. As the AI arms race heats up, China’s tech giants are doubling down on innovation to compete globally.
With Ernie 4.5, Baidu aims to reaffirm its position as a leader in AI development, betting that openness and advanced capabilities will drive broader adoption.