Baidu to Open Source AI Models After DeepSeek’s Success

Chinese tech giant Baidu reported its Q4 FY 24 earnings, with revenue declining 2% to 34.12 billion yuan due to weaker advertising sales. However, its Cloud AI unit grew 26% to 7.1 billion yuan, helping offset some losses. 

The company’s results exceeded analyst expectations of 33.32 billion yuan, highlighting the strength of its AI-driven initiatives despite economic headwinds and reduced ad spending.

During the earnings call, the company’s CEO, Robin Li, said that Hangzhou-based High-Flyer-owned DeepSeek’s open-source approach had accelerated AI adoption. 

"One thing we learned from DeepSeek is that open sourcing the best models can greatly help adoption. When the model is open source, people naturally want to try it out of curiosity, which helps drive broader adoption,” Li said. 

The company further announced that it will release the source code for its Ernie 4.5 models in June and offer premium chatbot services for free by April. “So our decision to open source ERNIE 4.5 series, is also backed by our deep confidence in our technology leadership, he added. 

Baidu was among the first in China to launch a ChatGPT-style chatbot in early 2023, with Ernie 4.0 claiming to rival OpenAI’s GPT-4. However, adoption has lagged due to strong competition, especially from DeepSeek’s R1 model, released last month.

Baidu’s AI Growth 

The company announced that in 2024, generative AI experienced strong market demand as enterprises increasingly adopted AI-driven solutions. Baidu’s AI solutions fuelled adoption and bolstered market confidence in its expertise.

“AI-generated content now appears in 22% of search result pages, with efforts focused on quality and relevance,” Li said during the earnings call.  

Baidu expanded beyond text generation to include short videos, nodes, agents, digital humans, and live streaming. The company placed strong emphasis on agents and said it envisions them becoming one of the most important forms of AI-native applications.